Skip to main content

Full text of "Analytical studies on the Bentiromide test for pancreatic function"

See other formats


ANALYTICAL STUDIES ON THE BENTIROMIDE TEST 
FOR PANCREATIC FUNCTION 



By 

H. THOMAS KARNES 



A DISSERTATION PRESENTED TO THE GRADUATE SCHOOL 
OF THE UNIVERSITY OF FLORIDA IN 
PARTIAL FULFILLMENT OF THE REQUIREMENTS 
FOR THE DEGREE OF DOCTOR OF PHILOSOPHY 



UNIVERSITY OF FLORIDA 
1984 



The author dedicates this dissertation to his wife, Susie, and their 
son, Jason, who have sacrificed much family time for this work to be 
completed. Without their patience, understanding and support this 
research would not have been possible. 



ACKNOWLEDGMENTS 

The author would like to express sincere thanks to Dr. Stephen G. 
Schulman for his guidance and support throughout this work. Thanks must 
also go to the members of the author's supervisory committee who 
provided many helpful discussions and encouragement during the course of 
this research. 

Appreciation is also extended to Dr. Phillip P. Toskes, Dr. Don 
Campbell and Cheryl Curington for providing clinical expertise and 
technical assistance. 



iii 



TABLE OF CONTENTS 



Pag 



ACKNOWLEDGMENTS 

ABSTRACT 

CHAPTER 

1 INTRODUCTION. 



11 



The Bentiroraide Test of Pancreatic Function 

Physiologic Factors Affecting the Bentiromide 

Test 

Analytical Factors Affecting the Bentiromide 

Test 1 

Room Temperature Phosphorimetry — Physical Aspects 1 

Room Temperature Phosphorimetry — Application 2 

High Performance Liquid Chromatography — Theory 21 

Bonded- Phase Liquid Chromatography Separations 4 

Ion-Pair Liquid Chromatography 51 

EXPERIMENTAL 5 

Apparatus 5 

Materials 5 

Synthesis of p-Acetamidohippuric Acid 6 

Bentiromide Administration and the Pancreatic 

Function Test 6 

Colorimetric Analysis of p-Aminobenzoic 

Acid in Urine 6 

Determination of p-Aminobenzoic Acid by 

Room Temperature Phosphorimetry 6 

Ion-Pair High Performance Liquid Chromatography 

of Bentiromide Metabolites 7 

Analysis of Bentiromide Metabolites by 

Room Temperature Phosphorimetry 7 

RESULTS 7 

Development and Optimization of the Room 

Temperature Phosphorimetric Method j 

Clinical Evaluation of the Room Temperature 

Phosphorescence Method q 



iv 



Development and Optimization of the Ion-Pair 
High Performance Liquid Chromatography 

Method 98 

Evaluation of Bentiromide Metabolism in 

Clinical Samples 125 

Analysis of Bentiromide Metabolites by Room 

Temperature Phosphorescence 140 

4 DISCUSSION 153 

Room Temperature Phosphorescence of p-Aminobenzoic 

Acid 153 

Variables Involved in the Chromatographic 

Technique 1 59 

Detection of Falsely Positive Bentiromide 

Test Results by Metabolite Analysis 164 

Physical Aspects of the Room Temperature Phosphorescence 

of p-Aminobenzoic Acid and Its Metabolites 168 

Conclusions 176 

APPENDICES 

I INTERFERING SUBSTANCES 177 

II CLINICAL EVALUATION 178 

III EXCLUSION CRITERIA 179 

REFERENCES 180 

BIOGRAPHICAL SKETCH 189 



v 



Abstract of Dissertation Presented to the Graduate School 
of the University of Florida in 
Partial Fulfillment of the Requirements 
for the Degree of Doctor of Philosophy 

ANALYTICAL STUDIES ON BENTIROMIDE 
FOR PANCREATIC FUNCTION 

By 

H. THOMAS KARNES 
December, 1984 

Chairman: Dr. S.G. Schulman 
Major Department: Pharmacy 

The urinary analysis of p-aminobenzoic acid which has been 

enzymatically cleaved from orally administered N- benzoyl- L-tyrosyl-p- 

aminobenzoic acid ( bentiromide) has been proposed as a screening test 

for pancreatic function. The test as it is currently carried out has 

certain drawbacks that are due to poor selectivity of the analytical 

method and false positive tests encountered with disease states other 

than pancreatic dysfunction. The phosphorescence of p-aminobenzoic acid 

at room temperature under various analytical conditions was studied in 

an effort to develop an analytical method that is superior to those in 

current use for evaluation of the bentiromide test. Room temperature 

phosphorimetry of p-aminobenzoic acid was then applied to the analysis 

of urine samples from subjects undergoing the bentiromide test and the 

procedure was found to be sufficiently accurate and precise for clinical 

evaluation and was more selective than colorimetric methods. 



vi 



An ion- pair high performance liquid chromatography procedure was 
also optimized for the analysis of p-aminobenzoic acid and its 
metabolites in urine so that metabolite concentration patterns could be 
correlated with disease states. It was found that patients with liver 
disease excreted significantly higher levels of the acetylated 
metabolite relative to the glycinated metabolite and this could be used 
to differentiate false positive bentiromide test results due to liver 
dysfunction. A room temperature phosphorimetric procedure for 
acetylated and glycinated metabolites of p-aminobenzoic acid was also 
developed but was shown to be of less diagnostic value than the 
chromatographic method. 

This study demonstrates the applicability of room temperature 
phosphorimetry and high performance liquid chromatography to enhancement 
of the selectivity of the bentiromide test. The results presented here 
also provide a better understanding of p-aminobenzoic acid metabolism 
and the effects of metabolic conjugation on its phosphorescence 
characteristics. 



vii 



CHAPTER I 
INTRODUCTION 

Orally administered bentiromide has been extensively studied as an 
indirect test of pancreatic function. Although the bentiromide test 
offers some advantages over other techniques, the method is relatively 
nonspecific with regard to the identification of pancreatic insuffi- 
ciency. The work detailed here explores various analytical techniques 
in an effort to eliminate nonspecif icity from both analytical and 
physiological sources. The analytical and clinical techniques employed 
in this work have been quite recently developed. Therefore, it is 
necessary to begin this dissertation with an overview of the bentiromide 
test and a discussion of pertinent theoretical concepts related to the 
analytical methods involved. 

The Bentiromide Test of Pancreatic Function 
The normally functioning pancreas is an efficient provider of 
amylolytic, proteolytic and lipolytic enzymes necessary for the 
digestion of food. Pancreatic secretion is mediated first by the vagus 
nerve in response to the sight, smell or taste of food (1). Addi- 
tionally, the gastric mucosa can stimulate pancreatic secretion by the 
stimulation of stretch receptors or by gastrin release. The intestinal 
phase which is the major contributor to pancreatic stimulation is 
mediated by the release of the two hormones secretin and cholecystokinin 
(2). Secretin stimulates the pancreatic secretion of water and electro- 
lytes whereas cholecystokinin acts to release the pancreatic enzymes. 



1 



2 



The major pancreatic enzymes are a-amylase, lipase, chymotrypsin and 
trypsin. 

Diseases of the pancreas are classified as inflammatory, traumatic, 
neoplastic or genetic. These diseases may vary widely as to the 
clinical picture and may present with or without exocrine pancreatic 
insufficiency (3). The general clinical features of pancreatic 
insufficiency are abdominal pain, diarrhea, and steatorrhea. 
Predisposing conditions such as cystic fibrosis or chronic alcoholism 
also provide valuable criteria for the identification of pancreatic 
disease. The diagnosis of pancreatic disease is difficult because of 
the anatomic location of the pancreas and a lack of adequate non- 
invasive diagnostic tests. The problem is compounded because certain 
clinical features and enzyme deficiencies may not be detected until 9Cf' 
of the pancreas has been damaged (4). 

Several tests of pancreatic function have been developed and can 
basically be categorized as direct or indirect tests. Direct tests of 
pancreatic function involve quantitative measurement of certain 
components of pancreatic secretions. These include the secretin and the 
secretin-cholecystokinin tests, both of which involve intravenous 
injection of pancreatic stimulants while monitoring the secretory 
response through aspiration of the duodenal fluid. Hormonal pancreatic 
stimulants vary greatly with respect to potency and the standardization 
of maximal normal secretory response has been difficult (5). 
Additionally, aspiration of duodenal fluid requires accurate placement 
of a nasogastric tube, usually by fluoroscopy, and results in radiation 
exposure (3). These procedures require hospitalization and some risk 
due to the invasive nature of such a technique. Although direct 



3 



pancreatic tests are considered the most reliable tests of pancreatic 
function, the disadvantages are obvious. 

It is desirable to screen patients on an outpatient basis and use 
the direct tests as confirmation only in selected cases. Relatively 
noninvasive screening is the goal in development of indirect tests of 
pancreatic function. Measurement of pancreatic enzymes in body fluids 
(blood and urine) can be used as an indicator of acute pancreatic 
damage, but is not a measure of pancreatic function and is of little 
value in the diagnosis of chronic pancreatic disease (6). The Lundh 
test (7) is considered a reliable indirect test of pancreatic exocrine 
function although it involves duodenal intubation. The test is based 
upon chymotryptic activity on the substrate N- 1 -benzoyl- 1 -arginine-ethyl 
ester after a controlled test meal has been given. The enzymatic 
activity measured is an indirect quantitation of the amount of 
endogenous cholecystokinin released in response to the test meal. The 
test does not require intravenous administration of hormone stimulators 
and provides a functional evaluation. False positive results have been 
noted in the presence of other diseases (8) and the required enzyme 
analysis is analytically difficult to deal with. 

The bentiromide test (9) is a relatively new approach to the 
indirect evaluation of pancreatic function. The test involves oral 
administration of the chymotrypsin-labile synthetic peptide K-ber.zoyl-1- 
tyrosyl-p-amino benzoic acid (bentiromide) shown in Figure 1-1. 
Digestion of this peptide by pancreatic chymotrypsin releases p- 
aminobenzoic acid (PABA) which can be subsequently absorbed, metabolized 
and excreted into the urine (9). The amount of PABA and its metabolites 
excreted in the urine, therefore, is an indirect measure of chymotrypsin 



4 



o 



CO — 



NH— CH- 
I 

CH 2 




( 

i 

■CO-j-M- 

i 

I 

i 

i 



OH 



Fig. 1-1. Structure of bentiromide showing the point of cleavage by 
chymotrypsin ( ). 



5 



activity and thus an indicator of exocrine pancreatic function. 
Preliminary studies by Iraondi and co-workers (10) suggested the test was 
a reliable indicator of pancreatic deficiency. They have subsequently 
shown that surgically induced pancreatic impairment can be accurately 
detected in a number of mammalian species (9). Since that time, the 
bentiromide test has been shown to be useful in determining pancreatic 
insufficiency induced by protein deficiency (11) and a number of 
clinical studies were carried out on human subjects (12-18). 
Bentiromide had been marketed for humans in Japan as early as 1980; 
however, the drug has just recently been approved for use in the United 
States. The peptide (bentiromide) was chosen on the basis of its 
sensitivity to hydrolysis by chymotrypsin and a general lack of 
hydrolysis by other pancreatic enzymes that may be present in the small 
intestine (9). The use of PABA provided an excellent marker molecule 
because it can be easily incorporated into peptides and is relatively 
nontoxic. It is also readily absorbed by the small bowel, rapidly 
excreted in the urine, and can be easily assayed. 

The bentiromide test is typically carried out by administration of 
a 500 mg oral dose of bentiromide and collection of urine for 6 hours. 
The urinary levels of PABA at different doses show proportional changes 
and suggest that there is little difference in test results between 
doses of 15 and 50 mg/kg (19). It is possible, however, that very high 
doses may cause a situation in which the amount of chymotrypsin becomes 
rate limiting (20). The standard dose is 500 mg and dose adjustment can 
be made at a level of 15 mg/kg for pediatric patients. The amount of 
PABA and its metabolites found in the urine is usually expressed as the 
percent of the dose recovered and a normal value is considered as 



6 



anything above 51% (19). This is based on a study of 61 normal subjects 
in which the mean recovery was found to be 71.7 ± 7. 2% and the cutoff 
was chosen to be the mean minus 2 standard deviations. Based on this 
cutoff point for normal values, the diagnostic sensitivity is 12% and 
specificity was determined to be 94.9$ when compared to a pancreatic 
diseased population. The defining concepts of "diagnostic" sensitivity 
and specificity are outlined in equations 1 and 2 below. 

c. Positive Tests (1) 
Sensitivity = v ' 

True Positives + False Negatives 

Specificity = Negative Tests (2) 

True Negatives + False Positives 

It is evident from the above relationships that sensitivity is a measure 
of persons with the disease that go undetected whereas specificity is a 
measure of those who do not have pancreatic insufficiency but whose 
tests are positive. Both criteria are very important because both the 
need to identify disease and the avoidance of subjecting normal 
individuals to more invasive tests must be considered. 

The sensitivity and specificity of the bentiromide test as it 
presently exists is not as good as those common to direct tests. The 
advantages of the bentiromide test lie in the practical utility of the 
method for screening since intravenous injections are not required, 
duodenal intubation is obviated, and there is no exposure to 
radiation. The analysis of PABA and metabolites in urine also 
eliminates the need for measurement of enzymes or unstable substances 
required by other methods. The analytical technique most commonly used 



7 



is the Smith modification of the Brat ton-Marshall method (21 ) which 
detects total urinary arylamine concentration. 

Physiological Factors Affecting the Bentiromide Test 

Although the bentiromide test seems quite attractive as a screening 
test for pancreatic function, there are several disadvantages that are 
not shared with other procedures. The test may not be useful in 
patients with severe hepatic disease since decreased metabolism could 
lower urinary PABA levels and thus create false positive test results 
(22,23). Additionally, patients with primary malabsorption in the small 
intestine may show false positive results due to decreased absorption of 
PABA (23). There has been some controversy over the degree of test 
specificity in patients with other diseases. With small bowel disease, 
patients may show symptoms similar to pancreatic insufficiency (24) that 
could cause difficulties with clinical diagnosis. This makes it 
especially important to be able to distinguish between the two disease 
states by diagnostic laboratory tests. Studies by Mitchell et al. (23) 
showed that 63$ of the patients tested with small bowel disease had low 
bentiromide test results, although all of these patients had a normal 
direct test for pancreatic function. This incidence of false positive 
tests is totally unacceptable in light of the clinical picture. 

Patients with small bowel disease and a normally functioning pan- 
creas can be detected, however, with other diagnostic tests. An oral 
dose of xylose (25 g) followed by urinary analysis of that amount, which 
was absorbed through the small bowel and excreted, is an effective 
indicator of small bowel function and is not subject to chymotrypsin 
activity (25). This test has been administered concomitantly with the 
bentiromide test in dogs and good results have been obtained (26). 



8 



Although concomitant administration in humans has not yet been 
established as a reliable means of differentiation, separate admini- 
stration of the two drugs has. Toskes and Greenberger (25) suggested a 
diagnostic algorithm for patients that present with symptoms typical of 
both small bowel and pancreatic disease. In this algorithm, patients 
with an abnormal bentiromide test and normal xylose test would be 
suspected of pancreatic insufficiency. Patients with abnormal results 
for both tests would have to undergo small bowel biopsy to assess small 
bowel function and patients with normal tests for both would be sub- 
jected to direct pancreatic function tests. With the use of this 
scheme, the authors proposed that the exact cause of symptoms could be 
determined in 85% of the cases. 

For the case of liver disease, animal studies have shown 
bentiromide is metabolized to p-aminobenzoic acid (PAEA) in the gut and 
subsequently conjugated by the liver to form both N-acetyl and glycine 
conjugates (Figure 1-2) which are the major metabolites found in urine 
(27,28). Glucuronide formation has also been observed for these 
derivatives (28), although this metabolic aspect may be poorly related 
to liver function since glucuronidase enzymes are present in many 
tissues (29). 

Studies have been carried out by Deiss and Cohen using PAHA synthe- 
sis as a model for liver function (50). In this work, various normal 
and pathological patients were given large doses (3 g) of PABA, and PAHA 
synthesis was monitored in serum samples. All pathological specimens 
tested demonstrated some negative deviation from the normal mean, 
although there was a great deal of variability depending on the type of 
liver disease. Studies have also shown that acetylation of PABA can be 



9 



COOH 




CONHCH 2 COOH 



T 

NH 2 




p-aminobenzoic acid (PABA) 



p-aminohippuric acid (PAHA) 



COOH 



CONHCHoCOOH 






HNCOCH. 



HNCOCH- 



p-acetamidobenzoic acid (PAABA) 



p-acetamidohippuric acid (PAAHA) 



v ig- 1-2. Structures of p-arcinobenzoic acid and its metabolites, 



10 



used as an indicator of liver function (31 )• There are, however, con- 
flicting reports and many factors affect acetylation of PABA in vivo 
(32). 

These studies show that PABA conjunction is decreased in patients 
with liver dysfunction and is dependent on the severity of the 
disease. It also follows that impaired conjugation may cause decreased 
urinary recovery of PABA metabolites resulting in false positive benti- 
romide test results. This was shown to be true in clinical studies 
involving liver patients and normal controls (22,23) where 50% of the 
liver patients studied had false low bentiromide test results. It also 
seems that conjugation impairment is dose related since the effect is 
more pronounced with high doses (1-2 g) of bentiromide. It appeared 
that PABA conjugation was impaired only in cases of severe, obvious 
liver disease and there are many commonly used laboratory tests to 
identify these patients. However, it would be advantageous to normalize 
the bentiromide test so that it could be used in patients with severe 
liver disease. 

These physiologic interferences with the bentiromide test have been 
adequately compensated for by the determination of a PABA excretion 
index (22). In this approach, an equivalent test dose of underivatized 
PABA is given several days prior to bentiromide administration. The 
urinary recovery from pure PABA is then compared with that of 
bentiromide through a ratio of the two results. This has been shown to 
correct for both liver and small bowel disease but has the disadvantage 
of requiring two tests be carried out at different times. 

Keasurement of PABA metabolites in plasma samples after bentiromide 
administration has also been shown to detect pancreatic insufficiency 



1 1 



(33)- This may also provide a means to avoid false positives due to 
liver dysfunction if unconjugated PABA alone were measured. Clinical 
studies involving liver patients, however, have not been done. 

False negative bentiromide test results may also arise from 
physiological factors (24). In a study by Toskes (19), 20% of patients 
with well defined pancreatic disease showed normal test results. The 
explanation of this is not clear although it has been suggested that 
variations in duodenal pH may have been responsible (34). Other 
studies, however, have failed to confirm these results (35). Other 
factors that may contribute to the observed false negative results 
include carryover effects from pancreatic enzyme therapy (36), non- 
pancreatic enzymes that can cleave bentiromide (37), and bacterial 
overgrowth in the small intestine (38). 

Analytical Factors Affecting the Bentiromide Test 

False negative results can also be obtained due to the lack of 
specificity of the analytical procedure (39). Concurrent use of certain 
drugs (see Appendix I) and even some food substances (40) can cause a 
false elevation of urinary arylamine concentrations and possibly a false 
negative test. The drugs listed in Appendix I must be discontinued 
three days prior to the administration of bentiromide and could result 
in significant interruptions in necessary drug therapy. It is also 
difficult to totally regulate a patient's intake and some substances 
which are not suspect may also raise urinary arylamine levels. 

Analysis of total PABA metabolites in urine from patients 
undergoing the bentiromide test have classically been carried out by 
colorimetric methods. The most commonly used method is the diazo- 
coupling method of Bratton and Marshall which was first used for 



12 



analysis of sulfanilamide (41 ) and subsequently modified by Smith 
(21). In this method, urine samples are hydrolyzed in 1.3 H HC1 to 
convert PABA metabolites to the parent compound. The hydrolysate is 
then diluted according to the urinary volume collected over a 0-6 hr 
interval so that sample concentrations are within the range of the 
standard curve. The diluted hydrolysate mixtures which contain 
liberated PABA are then reacted with sodium nitrite via its primary 
amine group to form the diazonium chloride. Ammonium sulfamate is added 
to scavenge excess sodium nitrite so that it will not react in the next 
step. Finally, N-( 1-Naphthyl) ethylenediamine dihydrochloride (UEDA) is 
added and reacts with the diazonium chloride to form the azo-dye 
chromophore. Absorbance is measured at 550 nm. A more convenient 
colorimetric procedure has been developed by Yamato and Kinoshita (42) 
which uses the chromogenic reagent p-dimethylaminocinnamaldehyde 
(DACA). This method has the advantages of a one step color reaction, 
more stable reagents and a more stable colored endproduct. Although 
this method is slightly more convenient, it is no more specific than the 
Bratton-Marshall procedure because both reagents react with all primary 
arylamines present in urine hydrolysates. The non-specificity of the 
colorimetric methods was demonstrated in a study where 2 of 59 fasting 
subjects excreted significant aromatic amines in their urine. 

Another possible source of error with these methods may arise from 
the acid hydrolysis procedure (43). Shosoki et al. (39) has shown that 
acid hydrolysis does not completely convert PAAHA (a major PABA 
metabolite) to the parent compound. Fortunately, PAAHA is primarily 
converted to PAHA which is a primary aromatic amine and therefore reacts 
to form a chromophore similar to that of the PABA-chromogen complex. 



13 

Differences in the absorptivities of these complexes and varible PABA 
metabolism could, however, lead to erratic results. The study by 
Shosoki et al. confirmed that alkaline hydrolysis completely converted 
all PABA metabolites and provides a better alternative since it is 
advantageous to measure a specific analyte rather than the additive 
contribution of two analytes. 

High performance liquid chromatography (HPLC) with electrochemical 
detection has been proposed as a more specific alternative to colori- 
metric procedures (39). In this method, alkaline hydrolysis was used 
and the results indicated good precision and accuracy. The method also 
compared well with the DACA colorimetric procedure. The selectivity of 
the method was reported to be superior to colorimetric methods although 
this was not thoroughly documented with drug interference studies. The 
HPLC method is potentially superior analytically although the procedure 
is more costly and time consuming than colorimetric analysis. 

Both physiological and analytical problems with the bentiromide 
test as it is presently carried out have been discussed in this 
section. The following sections will be dedicated to the establishment 
of a theoretical and practical treatment of the analytical methods 
involved in this research. Room temperature phosphorimetry will be used 
as an alternative to existing analytical methods in an effort to improve 
the selectivity of the test. In addition, concentration patterns of 
PABA metabolites will be studied by ion-pair HPLC so that physiological 
problems may be dealt with in this manner. 

Room Temperature Phosphorimetry — Physical Aspects 

The radiative transition from the lowest excited triplet state to 
the ground singlet state in organic molecules is called 



14 



phosphorescence. The triplet state is not normally accessible by direct 
excitation because electronic transitions from the ground singlet state 
to the triplet state are spin forbidden and occur with very low 
probability (44). The sequence of events that result in population of 
the triplet state is shown in Figure 1-3. If the absorption process 
(labeled A in Figure 1-3) is assumed to be a transition from the ground 
singlet state (Sq) to an excited vibrational level of an excited singlet 
state (S.,,S2), the molecule can lose the absorbed energy through several 
pathways. These pathways are governed by the kinetics of the competing 
processes. The vibrational energy is usually lost by transferring 
energy to neighboring molecules through vibrational relaxation (VR) to 
the lowest vibrational level of the electronic state in which it 
resides. The competing process, however, in which an excited vibra- 
tional level loses its energy by emitting a photon equal in energy to 
the difference between its existing state and the ground singlet state, 
occurs at a much slower rate and thus with a much lower probability. 
Once in its lowest vibrational level, the kinetically favored event is 
usually internal conversion (IC) to the lowest excited singlet state 
(S^). The molecule can then return to the ground state through internal 
conversion although this process is less probable than internal conver- 
sion from S2 to because of the greater energy separation between 
and Sq. The radiative loss of energy from an excited singlet to the 
ground state is called fluorescence (F) and the efficiency with which it 
competes with internal conversion to the ground state depends on the 
type of molecule and its environment. 

An alternate pathway to internal conversion and fluorescence is 
intersystem crossing (ISC) in which an electron changes the direction of 



1 U < 



n 

S 2 — r 




Pig. 1-3. Simplified energy level diagram showing singlet (S) and 

triplet (T) states for an organic molecule. Solid vertical 
lines represent absorption (a), fluorescence (F) and 
phosphorescence (P). Dotted lines indicate vibrational 
relaxation (VP.). Internal conversion (TC) and intersystem 
crossing (ISC) are represented by vertical and diagonal wavy 
lines, respectively. 



16 



its spin within the molecule to place it in an excited vibrational level 
of the triplet state (T^). Following intersystem crossing, vibrational 
relaxation to the lowest level of is very rapid. If the transition 
from the triplet state to the ground state is radiative, the process is 
called phosphorescence (P) whereas non- radiative transition again 
involves intersystem crossing. 

Both fluorescence and phosphorescence are analytically useful 
processes although phosphorescence can only be observed from a limited 
number of molecules. Phosphorescence, like fluorescence, is most likely 
to occur in aromatic molecules and their derivatives which have 
restricted vibrational freedom. 

Phosphorescence is generally distinguished from fluorescence on the 
basis of the longer decay-time of the former. Fluorescence decay-times 

_Q n 

are on the order of 10 3 to 10 seconds, whereas phosphorescence decay- 
times are generally between 10"^ and 10 seconds (45). The spin for- 
bidden nature of phosphorescence is responsible for the relatively long 
decay-times observed (46). Phosphorescence emission maxima also occur 
at lower energy than corresponding fluorescence maxima (47). Inter- 
system crossing occurs in molecules where the lowest excited triplet 
state lies slightly lower in energy than the lowest excited singlet 
state (48). 

The primary difficulty associated with analytical phosphorescence 
is that the triplet state, from which phosphorescence originates, is 
susceptible to radiationless decay by molecular collisions (44,49). 
This has led to the conduct of phosphorescence at very low temperatures 
in rigid matrices, in order to avoid these collisions. For many years, 
cryogenic spectroscopy was the only way to observe strong 



17 



phosphorescence from a variety of organic compounds (50-52). It has 
been shown more recently, however, that strong phosphorescence can be 
detected from a broad range of compounds at room temperature when 
rigidly adsorbed on a suitable support medium (44). Phosphorescence at 
room temperature has also been observed in solution when the emitting 
species are incorporated into stabilizing micelles which hinder the non- 
radiative deactivation pathways (53). These developments have opened up 
many new areas of interest in the application of phosphorimetry to 
chemical problems. They also provide additional insight into the 
phosphorescence phenomenon. 

Because of the long lifetime of the triplet state and the fact that 
collisional deactivation is very effective in bringing about 
radiationless decay in fluid media, phosphorescence is rarely observed 
in solution at room temperature. There have, however, been various 
attempts made to observe phosphorescence in solution at room 
temperature. In a limited number of cases, phosphorescence at room 
temperature has been observed from solutions that have been thoroughly 
degassed to remove small amounts of dissolved oxygen that quench the 
triplet state (54). Some compounds show phosphorescence in the gas 
phase due to a decrease in the frequency of molecular collisions at low 
pressures (55-58). A number of compounds exhibit strong phosphorescence 
at room temperature when embedded in rigid media such as glasses and 
plastics (59). In each of these cases, observed phosphorescence at room 
temperature has been either too weak for analytical purposes or there 
have been problems with sample preparation (60). 



18 



Substrate Interactions in Room Temperature Phosphorescence 

Many polar and ionic organic compounds have been found to exhibit 
phosphorescence at room temperature when adsorbed on materials such as 
silica, paper, alumina or cellulose (61-64). Since these early reports, 
a large number of compounds capable of phosphorescence in this manner 
have been identified. The phosphorescence of organic compounds adsorbed 
on solid supports is called room temperature phosphorescence (RTP). 

Studies by Schulman and Walling (64) showed that nonionic forms of 
compounds displayed very little RTP relative to their ionic counter- 
parts. These studies suggested that surface adsorption held the 
phosphors rigidly enough to limit collisional deactivation since 
adsorption of an ionic compound should be considerably stronger than 
that of the nonionic form. Subsequent studies have shown that reduction 
of the polarity of the adsorbant greatly decreases the intensity of RTP 
(64). It is now generally accepted that molecules capable of strong 
adsorption onto the support material prove to be the most efficient 
phosphors (44). 

Hydrogen bonding of the phosphor to the support material was also 
investigated as a means to restrict collisional deactivation of the 
phosphorescent compound (65). These experiments showed that reduction 
of the number of hydrogen bonding sites on the adsorbant produced a 
decrease in RTP intensity. Further studies supported the involvement of 
hydrogen bonding by comparing the RTP intensities of ortho and para- 
aminobenzoic acid adsorbed on sodium acetate (66). Para-aminobenzoic 
acid, which is strongly hydrogen bonded to the support according to IR 
data, demonstrated very intense RTP whereas o-aminobenzoic acid did not 
show RTP. The ortho isomer did not reveal strong hydrogen bonding to 



19 



the support, probably due to intramolecular hydrogen bonding of the 
substituents. The support-phosphor interaction is not yet fully 
understood and a number of anomalies exist which suggest the involvement 
of factors in addition to those already discussed. 
Influence of the Matrix on Room Temperature Phosphorescence 

There are many environmental effects on RTP intensities and 
investigations of these effects have led to a better understanding of 
phosphor-support interactions. It was noted by some initial investi- 
gators that thorough drying of compounds adsorbed on paper and other 
supports was necessary to observe RTP. The effects of moisture on the 
RTP of sodium 4-biphenylcarboxylate was investigated by measuring the 
RTP intensity as a function of relative humidity (65). The emission 
intensity was found to decrease dramatically with increasing relative 
humidity. The reduction in RTP intensity was attributed to displacement 
of the hydrogen-bound phosphor by water, since hydrogen bonding was 
considered the primary adsorption process. Other studies showed that 
RTP could be observed without drying if anhydrous solvents were used and 
the sample was protected from humid air (67). These results support the 
adsorption hypothesis since competition for hydrogen bonding sites by 
water would increase the mobility of the phosphor and, hence, the 
chances of collisional deactivation. 

Triplet state oxygen is a potent quenching agent of the excited 
triplet state through energy transfer. It is, therefore, surprising 
that strong RTP has been observed even under an atmosphere of pure 
oxygen. This suggests that the sample matrix somehow inhibits oxygen 
quenching (64). Subsequent studies have shown, however, that oxygen 
quenching does occur, to an extent depending on the sample matrix and 



20 



the phosphor (65,68). A possible explanation for this may be that some 
oxygen is trapped in the sample matrix when dried in the presence of 
oxygen since the more strongly the phosphor is adsorbed to the sub- 
strate, the greater the quenching effect of oxygen. Oxygen quenching of 
phosphorescence also appears to occur to a greater extent in humid 
atmospheres, which suggests that there is greater penetration of oxygen 
to the phosphor in the presence of water (67). 

The addition of external heavy atoms to the phosphorescence matrix 
has been commonly known to result in a decrease in fluorescence yield 
with a corresponding increase in the phosphorescence yield (49). Heavy 
atoms are necessary to observe RTP from certain nonpolar polynuclear 
aromatic hydrocarbons (68). These compounds should still be quite 
susceptible to collisional deactivation since strong adsorption between 
a nonpolar compound and a polar adsorbant would be unexpected. One 
possible explanation of this phenomenon might be the formation of polar 
■"-complexes between the compound and the heavy atom perturber that 
provide protection against collisional deactivation (68). The polar n- 
complexes may also interact more strongly with the substrate, which 
provides another possible mechanism for the increase in phosphorescence 
yield. 

The heavy atom effect has been attributed to increased spin-orbit 
coupling induced by the heavy atom (69-74). This makes population of 
the triplet state more likely through an increase in the singlet-triplet 
intersystem crossing rate and results in a decreased fluorescence yield 
and an increased phosphorescence yield. This also accounts for the 
observation that phosphorescence lifetimes are generally shortened in 
the presence of heavy atoms (44). Although radiative decay is usually 



21 



more strongly enhanced, the opposite effect has also been observed 
(7?). Sodium iodide has been successfully used to enhance RTP 
intensities either by preparing samples in iodide solutions or by 
spotting the solutions onto samples prior to drying. The increases in 
RTP intenstities for some compounds upon addition of sodium iodide were 
reported to be much too large to be explained totally by an increased 
intersystem crossing rate (75). This supports the polar ir-complex 
adsorption theory. 
Spectral Characteristics 

Spectral features observed in RTP are generally similar to those 
found at low temperatures (76). Room temperature phosphorescence 
spectra generally demonstrate lower resolution of vibrational structure 
than low temperature phosphorescence, due to increased vibrational free- 
dom at higher temperatures (77). Red shifts, typically less than 10 nm, 
are also observed in room temperature spectra when compared to corre- 
sponding low temperature spectra. L'-type delayed fluorescence is more 
readily observed at room temperature than at very low temperatures due 
to the lack of thermal energy to excite the triplet molecule back to the 
lowest excited singlet state at low temperatures (46). Only minor 
effects on spectral features are observed for the same compound adsorbed 
on a variety of support media (14). The pH of sample solutions can have 
a dramatic effect on RTP spectral properties (64,78). Although this has 
not been investigated, it is reasonable to assume that similar effects 
may be observed when changing from acidic to basic support media. 
Although sample matrices have little effect on RTP spectral positions, 
lifetimes and intenstities can be significantly affected (79, 80). In 
addition to the heavy atom effect, many other compounds such as 



22 



potassium hydroxide, boric acid, and sucrose have been found to affect 
either phosphorescence intensity or phosphorescence lifetime when added 
to the RTP matrix (44). 

Phosphorescence at Room Temperature in Solution 

Phosphorescence observed at room temperature in micelle solutions, 
provides an interesting aspect for further study. As mentioned pre- 
viously, limited phosphorescence in solution at room temperature has 
been observed but only after vigorous nitrogen purging to remove dis- 
solved oxygen. Some compounds which do not exhibit phosphorescence upon 
nitrogen purging alone, show significant phosphorescence in purged 
solutions containing certain detergents above their critical micelle 
concentration (38). A possible explanation of this effect is micellar 
stabilization of the triplet state (50). It is proposed that phos- 
phorescent molecules are shielded from collisional deactivating 
processes when they are within the protective environment of the 
micelle. These studies have been carried out in both aqueous and non- 
aqueous solutions (81,82). 

Studies of phosphorescence from micellar solutions have usually 
employed internal or external heavy atoms. The replacement of sodium by 
a monovalent heavy metal ion (e.g., Tl + ) in sodium lauryl sulfate 
micellar solutions dramatically increased phosphorescence intensities 
reaching a maximum at 10-20%' heavy metal ion replacement (82). Uo RTP 
was observed in aqueous micellar solutions of naphthalene, pyrene, or 
biphenyl in the absence of heavy atoms although sensitivities comparable 
to those attained in other phosphorescence methods were reported when 
the heavy metal was added. Analytical Figures of merit have been 
reported for a limited number of compounds in aqueous micellar solutions 



23 



and are reported to generally compare favorably to low temperature 
measurements (82). 

Room Temperature Phosphorimetry — Application 

There are several published sources of specific information on RTP 
spectral characteristics and analytical Figures of merit for a number of 
phosphorescent compounds on various support media (83,84,85). Data per- 
taining to compounds of interest in biochemistry, clinical chemistry, 
pharmaceutics and environmental chemistry have been reported. Limits of 
detection for these compounds generally range from subnanogram to sub- 
microgram quantities (85). Room temperature phosphorescence provides 
much greater selectivity than other luminescence techniques due to the 
fact that fewer compounds demonstrate RTP than fluoresce or phosphoresce 
at room temperatures (84). Room temperature phosphorescence obviates 
cumbersome techniques necessary to observe phosphorescence at low 
temperatures, and the capability for automation makes RTP a favorable 
alternative to low temperature phosphorimetry. 
Quantitative Applicability 

Equations developed by Winefordner describe a linear relationship 
between RTP intensity and concentration that departs from linearity at 
high concentrations much like low temperature phosphorescence (83). 
This effect, in practice, may occur at lower concentrations than 
predicted due to inner filter effects, triplet-triplet annihilation and 
various other factors (45). The quantitative capability of RTP has been 
demonstrated for a number of compounds and shows promise for practial 
utility with real samples. 

The choice of the proper compound and support for RTP is an 
important factor in the success of an analysis. Highly conjugated 
compounds with one or more aromatic rings are good candidates. The 



24 



compound should have a high phosphorescence quantum yield at low 
temperature and contain highly polar or ionic groups in order to have a 
high probability of showing RTF. As mentioned previously, certain 
nonpolar compounds exhibit RTP but only with the aid of a heavy atom 
perturber. 

Useful support materials usually contain either hydroxyl groups or 
numerous ionic sites (83). The most common support material has been 
paper. Silica gel and sodium acetate have also been widely used as RTP 
support materials. Selection of an optimum support has been largely 
trial and error, up to this point, and investigation is still in the 
early stages. 

Sample preparation in RTP usually involves application of the 
sample solution (3-5 ul) onto the support and thoroughly drying it. The 
drying step is especially critical due to the dramatic decrease in RT? 
intensities when the sample is exposed to moisture. Drying has been 
accomplished with heat lamps, blow dryers, dry nitrogen purge and 
laboratory ovens although some compounds may require maintenance under 
anhydrous atmospheres (83). 

For quantitative purposes, sample application volumes should be 
precisely measured and the amount of surface area covered should be 
controlled (86). Frying times should be optimized and kept constant. 
The presence of oxygen and temperature fluctuations (87) have been 
implicated as sources of imprecision in RTP and these variables should 
be controlled by drying under nitrogen at constant temperatures for best 
results. The pH of the analyte solution and support should be optimized 
to provide maximum RTP intensities. Some heavy atom perturbers have 
been shown to preferentially enhance RTP emission of individual 



25 



components in a multi-component mixture. This observation has been used 
to selectively determine each of the components in a multi-component 
system and has been termed selective external heavy atom perturbation 
(SEHAP)(88). 

Instrumentation for Room Temperature Phosphorimetry 

Although there is no commercially available instrumentation 
specifically designed for RTP detection, there have been several 
modifications of conventional luminescence instruments to accomodate the 
technique. Typically, the sample is mounted on a holder which allows 
positioning of the paper or other support material so that optimum 
excitation-emission geometry can be attained. The holder takes the 
place of the quartz Dewar flask used in low temperature measurements and 
the sample compartment is usually equipped with a gas purging system to 
avoid quenching by oxygen and humid air. The unit may also be 
temperature controlled to avoid possible imprecision as discussed 
earlier. The Aminco-Bowman spectrophosphorimeter has been reported to 
be suitable to receive these modifications (76). 

A modification of the Schoeffel spectrodensitometer which allowed 
viewing of RTP from several spots on one sheet of paper or thin layer 
chromatography plate was described by Ford and Hurtubise (89). The 
modified reflection assembly allowed the photomultiplier-source exit 
distance and photomultiplier angle to be varied to maximize reflected 
RTP reaching the detector system. These modifications were made to 
accommodate a rotating disc phosphoroscope. 

The most innovative instrumental design was constructed to 
demonstrate the automation capability of RTP (90). This system employed 
a filter paper guide that allowed continuous feeding of a paper tape 



26 



under a syringe for sample spotting, through a drying oven and finally 
into a sample compartment of a modified spectrophosphorimeter. Emission 
from the analyte spot was integrated as the paper tape moved at a 
continuous speed over the aperture. The semiautomated system was 
capable of measuring 7 to 8 samples per minute. 

Several other instrumental techniques have been described as means 
to improve RTP analysis. Ellipsoidal and parabolic mirrors have been 
employed to increase the collection efficiency of the luminescence from 
small volume samples which are typical of RTP (91 )• Synchronous 
scanning of excitation and emission monochromators (92) and derivative 
techniques (93) have both been used as means to improve the selectivity 
of the method. 

Advantages of Room Temperature Phosphorescence 

Sensitivity, selectivity, automation capability and avoidance of 
cryogenic techniques are all significant advantages of RTP analysis. 
Other advantages of RTP that warrant consideration are the small amounts 
of sample required and the possibility of direct spectral 
identification/quantification of separated analyte spots on paper and 
thin layer chromatographic plates. The linear dynamic range of RTP 
analytical curves and measurement precision has been reported as 
favorable in some instances (84). 

In contrast, there are also a number of limitations that are 
observed in RTP methods. The uniformity in thickness and porosity of 
available supports varies considerably from lot to lot causing problems 
with accuracy and precision. RTP is also more susceptible than low 
temperature procedures to scatter and background errors associated with 
solid surface instrumental measurements (85). Both of these limitations 



27 

could be minimized by production of substrates which are more suitable 
for RTP measurement and development of better solid surface instrumental 
designs. 

A major limitation of RTP as an analytical method lies in the 
phosphorescence background observed between 400 and 600 nm (94). 
Virtually every support that has been evaluated and shown to produce 
significant RTP signals also exhibits considerable background 
interference (84). Various pretreatments, experimental conditions, and 
instrumental adaptations have been examined as a means to lower this 
background with little success. Heavy atom perturbers must also be used 
with caution since they may have a similar enhancement effect on the 
background signal. Even with these high background levels, limits of 
detection for strongly phosphorescent compounds can be in the 
subnanogram range and the consequence of the background is that 
appropriate corrections for the blank must be made. 

Phosphorescence at room temperature arising from micellar 
stabilization might seem to be a viable alternative although studies are 
quite recent and have not been widely documented. The vast number of 
micellar systems available could provide excellent possibilities with 
respect to analytical applications. These possible advantages to the 
micellar stabilization technique, however, are overshadowed by the 
procedural difficulties. The very strict degassing requirements, 
concentration dependence, and susceptibility to impurities of the system 
create limitations to the analytical usefulness of the method. 

Quantitative analysis by RTP is a very sensitive and selective 
method for a number of biologically important compounds. Application of 
RTP to clinical analysis has hardly begun, probably due to a lack of 



28 



familiarity with the technique. The potential for both quantitative and 
qualitative analysis has been established, however, and RTP methods 
should find their way into practical analytical situations in the near 
future. 

High Performance Liquid Chromatography — Theory 
Chromatography is essentially a physical method of separation in 
which components are distributed between two phases, a stationary bed 
and a mobile liquid or gas flowing through this bed. Liquid 
chromatography refers to any chromatographic process in which the moving 
phase is a liquid. Open-bed liquid chromatography has been used 
routinely as a separation method for the past 40 years although original 
reports date back to 1903. The growth of liquid chromatography has 
occurred in periodic spurts corresponding to major innovations in 
techniques and instrumentation. 

Liquid chromatography can be divided into several subgroups based 
on the nature of the stationary phase and the separation process. In 
adsorption chromatography, the stationary bed is a solid on which 
separation is achieved through adsorption-desorption steps. Partition 
chromatography involves separation based on distribution between the 
liquid stationary phase and the mobile phase. Chromatographic 
separations based on ion-exchange principles have also been developed in 
which the species to be separated are charged and are retained by the 
oppositely charged stationary phase according to their relative charge 
densities. Separations based on molecular size are carried out with 
stationary phases having precisely controlled pore sizes. In these 
methods, the sample is simply sieved according to molecular size 
differences as it is washed through the stationary phase. Further 



29 



subdivisions of liquid chromatography exist although the main divisions 
are simply those of adsorption versus partition methods. In all types 
of chromatography, smaller stationary phase particles provide the most 
efficient separations due to increased total surface area. Smaller 
particle size, however, creates practical problems since the mobile 
phase moves very slowly through a dense particle bed. This has led to 
the conduct of liquid chromatographic separations at high pressures for 
the purpose of lowering analysis time while maintaining efficient 
separation. The method of applying high pressures to liquid 
chromatographic columns in a closed system has since been termed high 
performance liquid chromatography and has brought about a new era in 
chromatographic analysis. The following section describes some of the 
theory involved along with practical considerations pertinent to this 
dissertation. 
Migration and Retention 

The differential migration of compounds which serves as the basis 
for chromatographic separations is governed by equilibrium concepts. 
Due to the nature of the flowing system, however, it will be shown that 
complete equilibrium is valid only at the center of the migration zone 
of the compound of interest. If chromatographic migration is to be 
related to the equilibrium distribution of molecules between phases, we 
must refer to the center of the zone and have a rapid exchange of 
molecules between phases. 

Chromatographic separation of components in a mixture is achieved 
through differences in their equilibrium distribution (K) between two 
phases. If C g and C' m are the concentrations of a component in the 
stationary and mobile phases, respectively, then, 



30 



K = C s / C m (3). 

Migration is assumed to occur only when molecules reside in the mobile 
phase and each time a molecule affixes itself to the stationary phase 
its migration is interrupted. A zone of component molecules is then 
observed to migrate smoothly at some fraction (R) of the mobile phase 
velocity. Since migration can be viewed as equivalent to the 
probability of molecules existing in the mobile phase, the fraction of 
molecules in the mobile phase at equilibrium is equal to the same 
parameter R and denotes complete equilibrium. However, R (known as the 
retardation factor) is an average time fraction and zone spreading 
occurs as a result of fluctuations from this value. Therefore, R is 
applicable only at the zone center and molecules at the front and rear 
of the zone deviate from true equilibrium. The value R, having been 
established as equal to the equilibrium fraction of solute in the mobile 
phase, can be related to the equilibrium solute fraction (1-R) in the 
stationary phase. Thus, the term R/(1-R) is the ratio of the mass or 
number of moles of solute in the mobile phase to that in the stationary 
phase. This then is equal to the product of the solute concentration in 
the mobile phase and the volume of the mobile phase (C m V m ) divided by 
the corresponding term (C V ) for the stationary phase and 

R = C m V m (4). 

TT^Rl c s v s 

If the partition coefficient (K) is substituted from equation (3), we 
have 



31 



m 

TT^rT kv s 



V - (5). 



If equation (5) is solved for R, an expression equivalent to the classic 
equation of Martin and Synge (95) is produced: 



V m (6), 



m s 



Equation (6) has been derived as though there were a volume associated 
with the stationary phase, as is the case for the various classes of 
partition chromatography. However, for adsorption chromatography, V g 
can be replaced by the surface area of the stationary phase to attain an 
equivalent expression. If retention is due to both adsorption and 
partition mechanisms (e.g., partial adsorption in partition 
chromatography; Figure 1-4), equation (6) can be extended to the 
following (96): 

V m (7). 
V m + K i V si 

where and are the equilibrium constant and apparent volume of 
stationary phase which are experimentally derived and account for the 
adsorption process. At any rate, R can always be related to the 
equilibrium partition coefficient and the amount of each phase. Various 
forms of this equation exist, many of which can be used in 
chromatographic practice to predict the degree of retention for a given 
system. A similar parameter (k'), which is essentially the reciprocal 



32 



o 



Mobile phase 



o 

Stationary film 
£2- 




Stationary phase 
diffusion 

Absorptive transfer 



Adsorbed solute molecules 



Fig. 1-4. Idealized system for simultaneous adsorption and partition 

processes in liquid chromatography showing two mass transfer 
terms . 



33 



of R, is popular among chromatographers and is referred to as the 
capacity factor (97). 
Zone Spreading 

The effectiveness of chromatographic separation depends on two 
important factors: control of the relative migration rates and the 
amount of zone spreading. These two factors are equally important in 
attaining a desired separation since the separation is equally enhanced 
by either doubling the difference between the migration rates of 
different solutes or halving the spreading of peaks. The control of 
migration rates is accomplished by employing specific interactions 
(e.g., choice of the stationary and mobile phases) and is beyond the 
scope of this treatment. Therefore, only zone spreading will be 
considered here. The physical processes that give rise to zone 
spreading are present in all forms of chromatography and are of 
particular importance when dealing with nonlinear partition isotherms 
(Figure 1-5; 98). Exact linearity of partition isotherms is rarely 
attained in liquid chromatography and ideal conditions would result in 
each zone maintaining its original shape as it migrated through the 
column. 

There are three main contributions to zone spreading: eddy 
diffusion, longitudinal diffusion, and mass transfer. Eddy diffusion 
arises from the differences in mobile phase velocity attributed to 
different flow paths that solute molecules take through the particle bed 
and spreading is caused by faster movement of solute molecules thorugh 
wider paths relative to those in narrow paths. Longitudinal diffusion 
is the band broadening process that results from diffusion of solute 
molecules in the direction of mobile phase flow and becomes significant 



34 




Fig. 1-5. Three types of partition isotherms (C 1 is the total 

concentration in phase one and C2 is the total concentration 
in phase two). Curve a represents the ideal situation where 
the distribution ratio does not vary with relative 
concentrations. Curve b shows variations which would be 
observed for solute association. Curve c demonstrates the 
isotherm encountered frequently when phase one is an adsorbed 
phase. 



35 



only under conditions of stopped-f low and at low flow rates. Mass 
transfer spreading of the chromatographic band is primarily due to 
differences in the residence time of solute molecules on the stationary 
phase. Molecules that diffuse deeper into the stationary phase or are 
adsorbed more stongly spend more time in the stationary phase before 
they return to the mobile phase and will thus lag behind the main 
concentration of solute molecules. The relative importance of each of 
these contributions to zone spreading varies with the type of 
chromatographic system and conditions and is due mainly to the 
differences between physical properties of mobile phases. Mass transfer 
effects are the most significant; therefore, diffusion processes 
contributing to zone spreading will simply be described whereas mass 
transfer equilibrium will be considered in detail. 

The Theoretical Plate Concept . The theoretical plate concept is 
very useful in the description of zone spreading. This model was 
introduced into separation science to describe the distillation 
process. The efficiency of the distillation separation is determined by 
the number of condensation-volatilization steps required to attain a 
specific degree of purity. These hypothetical steps occurring during 
the repeated vapor-liquid equilibria were called theoretical plates. 
Understanding this process as it pertains to chromatography requires one 
to consider that the chromatographic column is divided into discrete 
segments corresponding to individual extraction tubes. Each segment 
contains an amount of mobile phase and stationary phase in which 
complete equilibration of solvent distribution has occurred. As a 
sample is added to the column, it enters the first such segment and 
solvent equilibration occurs. A fractional amount of solvent remains in 



36 



the mobile phase of the first segment to be passed on to the second and 
another mobile phase volume unit is added to the first stationary phase 
segment. As the process continues, a pattern of solute distribution 
develops which results in the observed separation. These segments are 
the theoretical plates and an entire chromatographic column contains a 
specified number (N) of these plates. A more detailed description of 
the theoretical plate theory is presented by Purnell (99). 

The theoretical plate model has two functions. First, it serves to 
describe chromatographic processes in the well defined terms of 
countercurrent distribution and, second, it provides a parameter H (the 
height equivalent to a theoretical plate) to characterize zone spreading 
and resolution. The use of this model to describe chromatographic 
processes is only approximate because the assumption that equilibrium is 
reached throughout the zone is not valid (100). However, the use of the 
parameter (H) is acceptable as a descriptive term to evaluate column 
efficiency. 

For given chromatographic conditions, the parameter H is 
approximately constant for different zones in the chromatogram and is, 
therefore, a measure of column efficiency. The parameter K is related 
to the number of theoretical plates (N) by the following expression: 

H = L / N (8) 

where L is the length of the column. Thus, H measures column efficiency 
per unit length and small values of H (large N values) correspond to 
more efficient columns. The parameter H can also be expressed directly 
in terms of zone spreading by 



37 



H = a 2 / L (9). 

The distance migrated by the zone center is represented by L and v 
corresponds to the standard deviation of the Gaussian curve which 
approximates zone distribution. The standard deviation of the Gaussian 
curve is a direct measure of zone spreading equivalent to roughly the 
quarter-width of the zone at the baseline. 

Having described zone spreading in terms of the theoretical plate 
height, we can now consider individual contributions to zone spreading 
in these terms. 

The plate height contribution due to longitudinal molecular 
diffusion (H^) is given by 

H L = 2 Y D n / V (10) 

where y is an obstruction factor that accounts for hindrance of 
diffusion by the column packing. The diffusion coefficient of the 
mobile phase is represented by D m and V is the mobile phase velocity. 

Giddings (100) has shown that the contribution to H from eddy 
diffusion is 

H E = 2 Ad ( n ) 

where d is the packing material particle diameter and ^ is a packing 
constant . 

Mass transfer and nonequilibrium . Mass transfer in the stationary 
phase contributes to zone spreading by two basically different 
mechanisms or by a combination of both. In adsorption chromatography, 
an abrupt molecular attachment or detachment process is the critical 



38 



step leading to sorption or desorption of solute molecules. Solvent 
molecules can detach from the stationary phase only if they possess 
sufficient energy to cause rearrangement or rupture of chemical or 
physical bonds. Partition chromatography, however, may be quite 
different in the mechanism of sorption-desorption in that penetration 
into and removal from the stationary phase are controlled by diffusion 
where a change occurs gradually. Mass transfer has been treated in 
terms of the random walk model (101,102) in which solute molecules are 
viewed as being displaced from the stationary phase in discrete forward 
or backward steps. The model is correct, however, only after a time 
period sufficient for solute molecules to form a Gaussian shaped zone 
profile has elapsed and should be reserved only to provide a basic 
understanding of the principal effects on zone spreading. 

A more accurate way of viewing the chromatographic process is that 
of a continuous flowing system. As a moving concentration pulse (solute 
zone) flows through a given column region, the concentration will 
increase to a maximum and then decrease back to zero. In the leading 
edge of the zone, a finite time is required to bring the stationary 
phase concentration to equilibrium with the continuously increasing 
concentration in the mobile phase. As long as the concentration 
continues to increase in the mobile phase, complete equilibrium of the 
stationary phase remains just out of reach with its concentration 
lagging slightly behind the mobile phase concentration (103,104). The 
reverse of this process is true for the trailing end of the zone since a 
new influx of mobile phase brings with it continuously more dilute 
solutions. This results in the stationary phase concentration being 
slightly higher than the equilibrium concentration in the mobile phase 



39 



after the zone center has passed. A schematic representation of this 
equilibrium lag in the stationary phase can he seen in Figure 1-6. The 
degree of equilibrium lag is dependent on the rate of mass transfer and 
will be slight if the rate of sorptive-desorptive exchange is large or 
if the zone migrates very slowly. This degree of nonequilibrium is very 
important since it determines the extent of zone spreading due to mass 
transfer and largely determines column efficiency. 

The nonequilibrium theory is based on changes in solute 
concentrations which result from the flow and kinetic processes observed 
in chromatography (105). Since the extent of departure from equilibrium 
is directly responsible for zone spreading, the concern is to quantify 
the effect in terms of equilibrium departure parameters: 

C m - C m* (12) 



and 



°s - C s* (13) 



e s = C s * 



where s and e c are equilibrium departure terms for the mobile and 

ill o 1 

stationary phases, respectively, and C g and C m denote concentrations per 
unit volume of packing material. The asterisk refers to equilibrium 
conditions at the zone center. From equations 12 and 13, it is evident 
that the e term is the fractional departure from eouilibrium. 

The equilibrium departure term can be related to column efficiency 
through the plate height parameter with the following relationships . 
The change in solute flux (aJ) is proportional to the overall 
concentration gradient as the mobile phase moves through any cross 



AC 




- — — Zone Center 

_______ Equilibrium Concentration 

Stationary Phase Concentration 

Mobile Phase Concentration 



Fig. 1-6. Diagram showing the stationary phase concentration lag (upper 
diagram) and mobile phase concentration forward displacement 
relative to the equilibrium zone center for liquid 
chromatography. 



41 



sectional area at a mean mobile phase velocity (v) and can be taken as 

AJ = C m * V (14). 

The flux can also be represented in terms analogous to diffusion, where 
D is the apparent diffusion coefficient for zone spreading as in the 
equation 

AJ = -D 8c/9z (15). 

The partial derivative 3c/3z represents the concentration gradient along 
the column axis (the partial form indicates that there is more than one 
variable and that all but one must be kept constant). Combining 
equations 14 and 15, 

- C m* £ m v (16). 

The plate height contribution is related to equation 16 by 

H = 2D/Rv (17). 

Therefore, substituting for D and expressing C m * as RC, we have 

H . - 2(RC)£ m (18). 
R 3 c /9z 

Simplification of equation 18 yields 



» = - 2e m (19). 
<*lnc/ a z 



This simple expression achieves the purpose of relating plate height to 
the equilibrium departure term. However, this does not mean that H is 
inversely proportional to 9lnc/3z or independent of v because the mobile 



42 



phase departure term will always be of a form capable of cancelling the 
31nc/ 3z term and introducing a term for v (see the following 
expression) : 

e = -(l-R)v 1 3 C r/ (20). 

m *a + k C * 3z 
d am 

where k & and are the first order rate constants for adsorption and 
desorption onto the mobile phase. If it is assumed that in partition 
chromatography the liquid stationary phase is a uniform film of depth d, 
the diffusional mass transfer coefficient in the stationary phase (D g ) 
can be similarly related to H by means of nonequilibrium concepts. The 
equilibrium departure term for diffusion controlled mass transfer 
through a liquid stationary phase can be expressed in terms of the 
following parameters: 

e = - Rv lnc Q-R)d' 2 (21). 
m Ds z 3 

Therefore, relation to plate height is simple since this parameter has 
already been defined for adsorptive processes in equation 19- A direct 
substitution of equation 21 into equation 19 with simplifications 
results in the relationship: 

H = 2/3 R(1-R) -lil (22). 
Cs 

Equation 22 is important since it shows that the plate height parameter 
(and thus column efficiency) is proportional to flow velocity, the 
square of the packing material film depth and the inverse of the 
diffusion coefficient in the stationary phase. These parameters can be 



43 



adjusted in practice and increased column efficiency is easily affected 
by the minimization of d' and the choice of stationary phase liquids in 
which D g is large. Flow velocity is not usually minimized to affect 
column efficiency because of the increased analysis time which 
results. Interestingly, equation 22 is the same as that derived for the 
partition case by random walk theory (101). The two values differ only 
by a numerical constant with the above expression being more correct. 
In practice, the numerical constant should be replaced by a variable 
term (usually q) which is called the configuration factor. 

The value of q depends on the shape of the pool of partitioning 
liquid (106) and the value of 2/3 derived by nonequilibrium theory is 
characteristic of a uniform film. 
Application of Liquid Chromatographic Theory 

In the previous section, the basic theoretical concepts of 
migration and zone spreading were discussed. The plate height parameter 
concept was described in terms of theoretical plate theory and related 
to the nonequilibrium model. The known complexity of chromatographic 
packing materials has led to many extensions and generalizations of the 
simple nonequilibrium theories discussed here (107-109). For example, 
complex mass transfer treatments can account for nonunif ormity of 
surfaces in adsorption chromatography. In addition, it should be 
obvious to a practiced chemist that Figure 1-4 should include mobile 
phase diffusion and interfacial transfer processes to be correct. If 
generalized nonequilibrium theory is applied, not only can experimental 
observations be interpreted in terms of theory, but specific information 
with regard to the degree of nonunif ormity versus column efficiency can 
be obtained. 



44 



Although we have seen that mass transfer terms constitute only a 
part of the total plate height, they are by far the most important from 
a practical standpoint and are the most interesting from a theoretical 
point of view. Plate height contributions due to other factors such as 
longitudinal diffusion and eddy diffusion in the mobile phase have been 
treated less rigorously throughout this introduction. Also, skew of the 
chromatographic migration zone (usually due to cases of slow 
equilibrium) has not been discussed. The attempt has been to describe 
the basic theoretical groundwork from which more complex treatments can 
precisely characterize chromatographic phenomena. 

Bonded-Phase Liquid Chromatography 

In this section, the practical aspects of liquid-liquid 
chromatography will be addressed with emphasis on the bonded stationary 
phase packings relevant to this work. Bonded-phase chromatography (BPC) 
is the most widely used means for high performance liquid 
chromatographic separation (110). Bonded-phase chromatography column 
packings are those with organic stationary phases chemically bonded to 
an inert support material. This is contrasted to classical liquid- 
liquid chromatography (LLC) stationary phases in which organic liquids 
are mechanically held to the support. Although the retention in BPC and 
LLC are governed primarily by partition mechanisms, many practical 
differences exist that permit wider application of BPC systems. 

The main advantage of BPC over LLC is that BPC columns are more 
stable because the stationary phase is chemically bonded and not easily 
removed during use. This obviates the necessity of column presaturation 
and allows the use of solvent gradients in the mobile phase without 
changing the composition of the stationary phase. A wider variety of 



45 



organic stationary phases (both polar and non-polar) can be used in BPC 
which allows both normal and reversed phase chromatography to be carried 
out. Reversed phase BPC, where the mobile phase is more polar than the 
stationary phase, is usually achieved by bonding hydrocarbons of various 
chain length to the stationary phase and using this in conjunction with 
organically modified aqueous mobile phases. The resulting separation is 
usually based on the molecular weights of analyte molecules and the 
types of functional groups present on them which affect their relative 
solubilities. Alkyl-substituted BPC packings provide the broadest 
utility of all packings for compounds with a molecular weight less than 
3000 Daltons, and should be the packing of choice for new analytes. 

Properties of the bonded-phase . Bonded-phase packings are prepared 
almost exclusively by attaching the bonded-phase to silica-based 
supports through reactions of surface silanol groups (111). These 
reactions result in silicate esters ( 1 12), silicon-carbon bonds (113), 
silicon-nitrogen bonds (114) and siloxanes (111) depending on the 
functional groups of the bonded-phase desired. The most widely used BPC 
packings are based on siloxanes (Si-O-Si-R) prepared by reacting silanol 
groups on the support with either organochlorosilane or organoalkoxy- 
silane. The bonded-phase can be either monomolecular or a polymerized 
multilayer coating. The stoichiometry of these reactions has been 
extensively studied (111) and the use of monomolecular or polymeric 
coatings depends on the type of support particle used. Low surface area 
pellicular supports may require polymerization of the reactants to 
achieve an adequate volume of bonded phase whereas monomolecular 
coatings are usually adequate for porous particles. 



46 



The degree of support material coverage depends on several factors, 
including the type of support (pellicular or porous) and the molecular 
volume and chain length of the bonded- phase modifier (115). The degree 
of coverage can readily be estimated by treatment with trimethylchloro- 
silane ("capping") which should not increase the carbon content of the 
packing or change its chromatographic characteristics if the surface had 
already been completely covered. 

Aside from the effect on support coverage, the chain length and 
structure of the bonded-alkyl group is the main factor in determining 
relative analyte retention. The log k' values for a given mobile phase 
increase linearly with hydrocarbon chain length in the absence of pore 
blocking effects by bulky bonded-phases (116). There also appears to be 
a linear relationship between log k' and the weight percent of organic 
coverage, independent of chain length (117). The sample loadability for 
reversed-phase BPC is also increased by increasing chain length although 
this relationship is nonlinear (116). 

The exact retention mechanism for BPC packings is not definitively 
established although several models have been proposed. Adsorption 
mechanisms have been proposed in which solvent molecules compete for 
sites on the organic surfaces of coated particles (118). The formation 
of an ordered liquid phase similar to a liquid crystal has also been 
suggested (119). Yet another mechanism has been proposed in which the 
organic coating interacts with molecules in the mobile phase creating a 
liquid phase which is defined by the coating (120). The assumption that 
BPC systems are equivalent to mechanically held liquid phases seems most 
appropriate for predicting retention behavior and should therefore be 
used for practical purposes. 



47 



Separation Variables 

The choice of solvent systems for BPC separations is generally 
similar to that for LLC. The stability of the bonded-phase is limited 
largely by the support material and most solvents can be used between 
the pH range of 2.0-8.5. Hydrocarbon BPC packings appear to be the most 
stable chemically although buffer and ion- pairing salts have a negative 
effect on their stability (121 ). Dissolved oxygen and strongly retained 
sample components also contribute to loss of column efficiency and 
appropriate degassing and sample clean up procedures should be used to 
avoid this effect. 

The primary factor determining the selectivity of reversed-phase 
BPC separations is the polarity of the solvent used. Some commonly used 
solvents and their relative polarities (P') are listed in Table 1-1. 
Methanol, acetonitrile and tetrahydrofuran are the most commonly used 
solvents in that order. Varying amounts of these organic solvents are 
generally added to water to adjust the strength of the mobile phase. 
The solvent strength decreases and k' values increase as the 
concentration and polarity of the solvent increases in reversed-phase 
BPC. Significant differences can be observed in separation selectivity 
by the replacement of one organic solvent with another, which sometimes 
results in a reversal of analyte elution order (122). Methanol, for 
example, can be used to decrease the strength of an aqueous phase 
without affecting the selectivity of water since both are proton donors 
and acceptors. Ternary mobile phases, such as methanol/dioxane/water, 
have also been used to provide unique selectivity for some compounds 
(123). 



48 



The pH of the mobile phase can greatly change the separation 
selectivity for ionizable analytes. Variations of pH are generally not 
effective for solutes that do not ionize. The retention time for a weak 
uncharged acid will be relatively constant with an increase in pH until 
the pH approaches the pKa of the acid. The retention time then 
decreases and levels off when the acid has become fully ionized. This 
is expected since only the nonionized form should partition into the 
hydrophobic stationary phase significantly. In contrast, the retention 
time of weak uncharged bases tends to increase as pH increases. 
Therefore, variation of mobile phase pH can be used effectively to 
separate structurally similar compounds with significantly different pK 
values . 

Temperature and flow rate are somewhat less useful in determining 
reversed phase chromatographic selectivity (124,125). Generally, 
increasing flow rate or temperature causes a linear decrease in 
retention for solute molecules. Temperature changes can decrease mobile 
phase viscosity, increase solute solubility, and alter separation 
selectivity to some degree although adjustment of the flow rate may be 
useful only for providing rapid separation. Decreasing mobile phase 
viscosity improves separation efficiency and BPC packings are typically 
stable to 80°C. BPC separations are usually carried out at room 
temperature with low viscosity solvents, except that operation at higher 
temperatures may be advantageous for relatively viscous mobile phases. 
Temperature effects on separation selectivity are generally useful for 
solutes of different functionality although these effects are not very 
pronounced and are unpredictable. Therefore, temperature adjustment is 
not usually done as an initial means of adjusting selectivity. 



49 



Achieving a bonded-phase chromatographic separation 

A reversed phase BPC (bonded-hydrocarbon) column should be selected 
first for use in a particular separation because of its broad 
applicability. The C 1Q packings offer the best characteristics for 
applications involving compounds with high water solubility and shorter 
chain columns (C^) are more appropriate for strongly retained 
hydrophobic molecules. Using a shorter chain hydrocarbon as the bonded- 
phase may also improve selectivity due to the higher mobile-phase water 
concentrations needed for these columns. Packings with intermediate 
hydrocarbon chain length (i.e., C Q ) represent a good compromise and are 
useful for the widest variety of samples. Drug analysis usually 
involves compounds of high water solubility and, therefore, C 18 columns 
are generally the most useful for these applications. 

The first choice of mobile phase is either methanol/water or 
acetonitrile/water and the optimum composition is usually found 
empirically. A good starting point is 1:1 organic modifier/water, and 
concentration adjustment should be made according to the elution of 
sample components. A different solvent should be used primarily to 
alter selectivity since different concentrations can effectively adjust 
solvent strength. 

The pH of the mobile phase is chosen either for selectivity or the 
suppression of ionization by protonation or deprotonation. Selectivity 
can be drastically affected throughout the pH range in which components 
are partially ionized. This partial ionization can also lead to band 
tailing so buffering sample components in the nonionized form is a 
better alternative if adequate separation can be achieved. 



50 



Gradient elution systems which normally increase the concentration 
of organic modifier with time are useful for samples with widely 
different components. Qualitative screening information can also be 
obtained through gradient elution. 

The temperature chosen for reversed phase separations are generally 
more efficient between 50 and 60°C which provides about twice the number 
of theoretical plates than that obtained at ambient temperatures. 
Column stability is greater at ambient temperatures, however, and higher 
temperatures should be avoided when possible. 

The optimum mobile phase flow rate is usually between 1-2 nL/min 
but depends on the type of mobile phase, the type and size of packing 
material, the operating temperature and the internal diameter of the 
column. Generally, the best flow rate is one in which the components 
elute quickly without losing resolution and the back pressure is 
maintained below 4,500 pounds per square inch. 

The amount of sample injected should be less than 50 Ug for 
pellicular BPC packings and less than 500 Ug for porous BPC packings. 
Sample volume should not exceed one-third the retention volume of the 
first peak of interest although larger volumes may be tolerated for 
injection solutions of lower solvent strength. 

Ion-Pair Liquid Chromatography 

Ion-pairing has been used in liquid-liquid extraction procedures 
for many years, although its application to HPLC has been rather 
recent. The advantages of ion-pair chromatography (IPC) arise from 
comparison with ion-exchange methods. Ion-exchange HPLC columns are 
less efficient, less reproducible, less stable and available in less 
variety than other types of columns. The limited column variety results 



51 



in a poor potential for selectivity in ion-exchange as compared with IPC 
methods. These advantages have been described by Schill and coworkers 
(126) who have contributed greatly to the understanding and acceptance 
of the method. 

In IPC separations, the stationary phase usually consists of the 
silanized silica packing normally used in reversed- phase BPC chromato- 
graphy. The mobile phase consists of an aqueous buffer, organic modi- 
fier and an added counter-ion of opposite charge to the ionized sample 
molecule. The hydrophobic character of these ion-pairs causes the ion- 
pair to be more strongly retained than the free sample ion and better 
resolved from hydrophylic interfering substances in the sample ( 1 27 ) . 

The simplest case of IPC can be described by assuming that the 
sample and pairing ion are soluble only in the mobile phase, whereas the 
ion-pair formed is soluble only in organic stationary phase. This would 
result in a distribution between the phases described in equation (23): 

RC00~ aq + IP + aq = RC00- IP + org (23) 

where RCOO" represents the analyte ion and IP + corresponds to the 
pairing ion. The subscripts "aq" and "org" refer to the aqueous and 
organic phases, respectively. The extraction constant E can then be 
defined in terms of concentrations of species as follows: 



E 



[RC00-IP+] org (24)> 



[RC00-] aq [iP + ] aq 



The extraction constant for a particular system depends on the 
temperature and composition of the mobile phase and can be related to 
the capacity factor k' as follows: 



52 



Ic' E[lP + ] aq (») 

m 

where V g and V m are the retention volumes of the stationary and mobile 
phases. In the absence of other effects, the retention of singly 
charged sample compounds is proportional to the concentration of the 
ion-pairing agent (128). The variation of [lP + ], therefore, provides 
additional means of controlling apparent solvent strength and can be 
used for differential elution of sample components. 

In addition to the ion-pair concentration parameter, separation 
selectivity can be changed by varying the pH. Also, IP concentration 
gradients can be employed to further control selectivity and analysis 
time. Positively charged sample compounds can be similarly influenced 
with negatively charged pairing ions and amphiprotic molecules can be 
paired with either negative or positive pairing agents. This great 
diversity of IPC techniques combined with the separation variables 
already described for bonded-phase HPLC makes it possible to tailor a 
separation for optimum results. The application of IPC to the control 
of separation has been extensively described (126-133) and this 
introduction discusses general considerations without reference to 
specific applications. 

Considerations Involved in Ion-Pair Separations 

Some investigators have proposed that the pairing-ion is adsorbed 
onto the organic stationary phase causing retention of sample ions 
through ion exchange mechanisms (128). This suggests that the simple 
model proposed in equation (23) is not completely adequate to describe 
the processes involved, although the simple model is sufficient to 



53 



predict separation parameters for practical purposes. Adsorption of 
ion-pairing agents onto the stationary phase may also change the 
separation characteristics of bonded-phase packing materials by adding 
to the apparent hydrocarbon content of the stationary phase. The 
practical consequence of both of these phenomena is that columns must be 
thoroughly flushed with counter-ion containing mobile phase so that an 
equilibrium coating can be achieved. Very large counter-ions are more 
susceptible to adsorption and must, therefore, be equilibrated for a 
longer time. 

Another problem in the use of ion-pairing agents is the need for 
buffer salts to control the pH of the system. This can reduce the 
formation of ion pairs, and consequently solute k' values, through a 
competing secondary equilibrium between the buffer ion and the pairing- 
ions (134). Those buffer ions with the same charge (cationic or 
anionic) as the sample ions have the largest effect. The effect also 
varies with the type of buffer ion. In an IPC separation of sample 
anions, k' values were decreased by the effect of secondary ions in the 
sequence NO^" > Br" > CI" > S0 4 ~ 2 ( 1 33) . Traditional buffers such as 
citrate and phosphate have been used for IPC separations although they 
also affect k' values. The ion-pairing agent itself provides an 
adequate buffer in some cases. 

The relative efficacy of different ion-pairing agents in increasing 
k' values depends primarily on two factors, the formation constant of 
the ion pair and the size of the counter-ion. The counter-ion 
functional group can have a significant effect on the ion-pair formation 
constant. Octane sulfate has been shown to be less efffective in ion 
pair formation than octane sulfonate for increasing the retention of 



54 



cations in reversed phase HPLC (135). Similar effects can be expected 

for different anionic pairing agents although tetraalkylammonium ions 

have been used almost exclusively. As expected, larger counter-ion 

molecules result in larger k' values in reversed phase IPC and k' values 

5 

have been increased 10 -fold when the counter-ion was varied from 
tetraethylammonium to tetrapentylammonium (130). The increase in log k' 
with total carbon number is linear although structural changes in the 
counter-ion such as those for surface acting pairing-ions will cause 
deviations from linearity. This is due partially to the concomitant 
ion-exchange and stationary phase altering mechanisms mentioned 
previously. The ion-pair formation constant may also be affected by the 
structure of the pairing-ion. 

Solvent strength or polarity factor (P') can be adjusted in IPC by 
the use of organic modifiers (i.e., methanol and acetonitrile) in the 
mobile phase although solvent polarity does not follow the normal 
sequence illustrated in Table 1-1. This is because solvent strength in 



Table 1-1. Solvent strength in reversed-phase BPC. 



Solvent p' 



Water 10.2 

Dimethyl sulfoxide 7.2 

Ethylene glycol 6.9 

Acetonitrile 5.8 

Methanol 5.1 

Acetone 5.1 

Dioxane 4.8 

Ethanol 4.3 

Tetrahydrofuran 4.0 

1-Propanol 3.9 



5^ 



IPC is also a function of the solvent's ability to stabilize ion pairs 
whereas solvent strength in BPC is mainly due to the solubility of polar 
nonionic solute molecules. The solvent strength in IPC should, 
therefore, be a function of both solvent polarity (P) and solvent 
dielectric strength (e) and can be predicted by the solvent function (P 
+ 0.25 e). 

Capacity factors are greatly affected in IPC by changes in pH and 
are usually accompanied by large changes in separation selectivity. 
Maximum k* values in reversed phase IPC are observed at pH values where 
sample compounds are completely ionized. However, organic acids that 
are retained equally for example at pH 3 may be readily separated at pK 
4 due to differences in their pKa values. Thus, pH variation offers a 
powerful tool for changing separation selectivity in IPC although this 
method is not appropriate for simple solvent strength adjustment. 
Capacity factor versus pH curves in reversed phase IPC will show the 
typical sigmoid relationship if the unionized sample ions are not 
retained by the stationary phase but the curves become more complex if 
this is not the case (133). 

Temperature and flow rate variations follow the same general trends 
discussed for reversed phase BPC. The effect of temperature, however, 
is somewhat more important since mobile phases used in IPC are more 
viscous and changes in selectivity appear to be more pronounced ( 1 33) . 
Therefore, temperature can be an important variable for optimizing 
separation selectivity in some IPC applications. 

Aside from the special considerations discussed here, the design of 
an IPC separation should proceed in the same manner as that for BPC. 
Special precautions must sometimes be taken to prevent band tailing due 



56 



to the dissociation of ion pairs in the organic phase. This can be 
overcome by either increasing the counter-ion concentration or choosing 
another IPC system. An added benefit of IPC is that UV-absorbing 
counter-ions can be used to increase detectability of some compounds by 
allowing sample bands to leave the column as detectable ion-pairs. This 
may obviate the need for derivatization and both the picrate (136) and 
2-naphthylsulfonate (137) ions have been used for this purpose. 

A brief description of the basic analytical and clinical aspects 
involved in the following research has been presented here. It should 
be apparent from this introduction that RTP analysis of urinary PABA 
from patients receiving bentiromide may provide more analytically 
selective test. Limits of detection for PAEA by RTP have been observed 
in the picomolar range (138) and RTP has been used to quantify PABA in 
vitamin tablets (139). The method is potentially adaptable to analysis 
of biological fluids and should prove to be more convenient than 
existing methods. 

The potential also exists to differentiate physiological non- 
specificity observed in the bentiromide test by establishment of PABA 
metabolite concentration patterns. If either hippurate synthesis or 
acetylating capacity is deficient in liver patients, false positive 
tests due to impaired liver function may be detected. 

These are goals of the following research and attempts will also be 
made to incorporate PABA metabolite differentiation into the RTP 
analytical format. A study involving the structural dependence of RTP 
signals from the various PABA metabolites will also be undertaken. 



CHAPTER 2 
EXPERIMENTAL 

Apparatus 

Room temperature phosphorescence measurements were made with a 
modified spectrophotof luorimeter (Aminco Bowman model SPF100) equipped 
with a rotating can phosphoroscope (140), high voltage ratio photometer 
(American Instrument Co., Silver Springs, MB) and a Hamamatsu IP21 
photomultiplier tube (Whatman, MA). A laboratory constructed multiple 
solid sample bar described by Ward et al. (141 ) was used with a modified 
sample compartment lid, and a 150 W xenon arc lamp provided excitation 
energy. An X-Y recorder (Allen Batagraph, Inc., model 715, Salem, NH) 
was used to record RTP spectra, and a strip chart recorder (Model SRG, 
Sargent-Welch Scientific Co., Skokie, IL) was used to record dr. ing 
histograms. Heavy atom and analyte solutions for RTP were spotted by 
use of an SMI adjustable volume (1-5 yL) micropetter (Emeryville, CA), 
and a dry heating block (Lab-line Instruments Inc., Melrose Park, IL) 
was used for alkaline hydrolysis and enzymatic incubation. Acid 
hydrolysis was carried out in a boiling water bath heated by a Corning 
model PC351 hot plate stirrer (Corning Class Works, Corning, NY) 
laboratory warmer. 

The pH measurements were made with a Markson pH meter (l-iarkson 
Science Inc., Bel Mar, CA) and ultraviolet spectra of analyte compounds 
were obtained using a Beckman model 25 spectrophotometer (Beckman 
Instruments, Irvine, CA). Visible spectra and analytical absorbance 



^7 



58 



measurements for colorimetric analysis were also made on the Beckman 
spectrophotometer. 

A Waters Model M-6000 A liquid chromatograph (Milford, MA) equipped 
with a Rheodyne 7 105 loop injector and 254 nm fixed wavelength detector 
was used for chromatographic analysis. Chromatography was performed on 
a Waters \i Bondapak (30x0.4 cm, 10 ym) column and an Omniscribe model 
3532 strip chart recorder (Houston Instruments, Austin, TX) was used. 
The column temperature was controlled with an Altech water jacket 
(Ceerfield, IL) connected to a Temptrol 15? (Precision Scientific, 
Chicago, IL) circulating water bath. 

Materials 

"Nanopure" de-ionized water (Barnstead system of Sybron Co., 
Boston, MA) and absolute ethanol (U.S. Industrial Chemical Co., New 
York, NY) were used for RTF analyte dilution. Para-aminobenzoic acid, 
p-aminohippuric acid (PAH) and p-acetamidobenzoic acid (PAABA) were all 
purchased from Sigma Chemical Co. (St. Louis, MO). Para-acetamido- 
hippuric acid (PAHA) was synthesized as described in the following 
section. Diethylaminocellulose (DE-81) anion-exchange filter paper 
(Whatman Chemical Separation Inc., Clifton, NJ) and S&S 903 (Scheilcher 
and Schuell Inc., Keene, NH ) were used as RTP solid-support materials. 
The S<5S 903 filter paper was treated with diethylene triamine 
pentaacetic acid (DTPA) obtained from Sigma Chemical Co. as previously 
described (142). Silver nitrate (Mallinckrodt Chemical Works, St. 
Louis, MO), potassium iodide (Fisher) and thallium (l) nitrate (PCR 
Inc., Gainesville, FL) were used as heavy atom perturbers. Anhydrous 
monobasic potassium phosphate (Mallinckrodt Chemical Works, St. Louis, 
MO), sodium hydroxide, phosphoric acid and sulfuric acid (all obtained 



59 



from Fisher Scientific Co., Fair Lawn, NJ) were also used in RT? 
analysis . 

Analytical reagent grade ammonium sulfamate (Eastman Chemical Co., 
Rochester, NY), N-( 1-napthyl)ethylenediamine dihydrochloride (Eastman), 
concentrated hydrochloric acid (Fisher) and sodium nitrite (j.T. Baker 
Chemical Co. , Phillipsburg, NJ) were used as reagents in colorimetric 
analysis . 

Methanol and acetonitrile were used as organic modifiers in 
chromatographic analysis and were HPLC grade from Fisher Scientific Co. 
(Fair Lawn, NJ). The water used for KPLC mobile phases was distilled 
and deionized by a Watts model M (Lawrence, MA) water purifier. 
Cationic ion-pairing agents ( tetramethylammonium chloride, tetraethyl- 
ammonium chloride, tetrabutylammonium chloride and hexadecyltri- 
methylammonium bromide) were all 95$ pure from Fisher Scientific Co. 
(Fair Lawn, NJ) and used without further purification. The 8- 
glucuronidase used for pretreatment of urine samples from clinical 
subjects was obtained from Sigma Chemical Co. (St. Louis, MO) and 
contained 570,000 "Fishman" units/gm of solid. A "Fishman" unit is 
defined as that quantity of enzyme that will liberate 1.0 v g of 
phenophthalein from phenolphthalein glucuronide per hour at 37 °C. 

Hexane sodium sulfate, 1 -octanesulf onic acid, octane sodium 
sulfate, decane sodium sulfate, and dodecane sodium sulfate (all from 
Eastman Chemical Co., Rochester, NY) were used as anionic pairing 
agents. All other chemicals used in chromatographic analysis (monobasic 
potassium phosphate, phosphoric acid, sodium acetate, acetic acid, and 
sodium hydroxide) were analytical grade from Fisher Scientific Co. 



60 



Synthesis of p-Acetamidohippuric Acid 
Para-acetamidohippuric acid was synthesized by the reaction of 
acetic anhydride (Fisher Scientific Co.) with p-aminohippuric acid as 
described by Vogel ( 1 43) - The reaction is diagrammed in Figure 2-1 and 
proceeds by a mechanism in which the amine group on PAHA 
nucleophilically attacks a carbonyl carbon atom on acetic anhydride. 
Half of the anhydride appears in the acyl product and the other half 
forms a carboxylic acid which is dissociated at the pH of the reaction 
mixture. 

The reaction was carried out by the addition of 5.0 g of PAHA to a 
mixture of 60 mL H 2 and 12.5 mL of concentrated hydrochloric acid. 
Three milliliters of acetic anhydride along with 3-85 g of sodium 
acetate were dissolved in 1 2 mL of H 2 and added to the mixture. The 
combination of sodium acetate and HC1 provided an initial pH of 4.2 
which kept PAHA primarily in its anionic form to prevent reaction of the 
carboxyl group with acetic anhydride. This mixture was allowed to react 
at 25°C while stirring until the precipitation of white crystals which 
corresponded to acetylated PAHA was complete (about 2-3 minutes). The 
crystals were filtered and washed with cold water. Purity of the 
product was evaluated by thin layer chromatography on silica plates with 
Acetone/Ethanol/water (50/38/12) as the solvent. The R f value for the 
product was 0.84 whereas PAHA demonstrated an R f value of 0.72. There 
were no impurities visualized in either the product or starting 
material. No additional peaks corresponding to unreacted PAHA or other 
reaction products were found by HPLC analysis which provided further 
evidence of product purity. Melting points obtained from the starting 
material and the product were 204-206 and 236-238°C, respectively, and 



61 



cj 
■- 

U 
3 
p, 
ft 

•H 

4? 
C 

•H 
£ 

■H 

CD 
O 

a 



i 



O 
i 



O O: 


o 


CO 


1 

o 




rO 




X 


X 




CJ 


+ 


X 









E 

c 

c 
-(-> 

a 

T3 
•rt 

f- 
TT 

>- 
— 

c 

CD 



o = o- 



\\ // 



-t-> 

CP 

o 



o = 



o = 



+J 

•H 

■c 

•H 

C 
CO 

O 
•H 
C 
N 
G 
<D 

c 

•H 
B 
.-; 
+j 
o 
o 

vv 

I 



'o 
o 
o 

CJ 
X 

o 

X 
2 

O =o - 




Cm 
C 



C 

c 



o 
CO 
CD 

U . 

■a 
a -h 

Eh cfl 



62 



the PAHA melting point was within 5° of a published result (144). Both 
NMR and IR spectra demonstrated peaks characteristic of the acetylated 
derivative ( 1 45 ) and were compared to spectra obtained from the starting 
material. 

Bentiromide Administration and the Pancreatic Function Test 
Subject Selection Criteria 

Normal adult subjects (18 years of age or older) were determined to 
be in good physical health by the clinical evaluation criteria listed in 
Appendix II. If any of the clinical evaluation criteria indicated that 
disease was present, the disease must not interfere with the absorption, 
metabolism or excretion of PABA and be controlled by medications other 
than those listed in Appendix I. Normal subjects were also excluded 
from the study if any of the exclusion criteria listed in Appendix III 
were met. 

Study subjects with chronic pancreatitis and exocrine pancretic 
insufficiency were identified by abnormal results for the 
secretin/cholecystokinin (CCK) stimulation test. To be considered 
abnormal, the maximum concentration of bicarbonate in aspirated duodenal 
fluid must not exceed 30 meq/L following stimulation of pancreatic 
secretion with 1 Ug/kg of intravenous secretin and 0.02 Ug/kg of 
intravenous cholecystokinin (146). Patients with small bowel 
malabsorption were diagnosed by an abnormal urinary d-xylose test and 
small bowel biopsy in some cases. Abnormal urinary d-xylose excretion 
was taken to be less than 5 g/5 hr after administration of a 25 g oral 
dose (147). Patients with liver disease were chosen on the basis of 
abnormal liver function tests and were evaluated to identify their 
underlying liver disorder by liver biopsy. These patients were not 



63 



evaluated by the CCK or d-xylose tests and can not necessarily be 
considered to be free of small bowel or pancreatic dysfunction. 
Bentiromide Administration and Urine Collection 

To each of 24 healthy adult volunteers, 5 patients with small bowel 
malabsorption, 6 patients with chronic pancreatitis, and 3 patients with 
exocrine pancreatic insufficiency, 500 mg of bentiromide (Adria Labs., 
Inc., Columbus, OH) were orally administered in 250 mL of water after an 
overnight fast. Just before dosing, the subjects were instructed to 
empty their bladders and consume 500 mL of water. An additional 500 mL 
of water was given and subjects remained fasting until the test was 
completed. Urine was collected and pooled for 6 h after drug 
administration, the total volume was measured and the specimen was 
mixed, divided into aliquots and stored at -4°C until analysis by both 
colorimetric and RTP methods. 

An additional 5 healthy adult volunteers (4 male, 1 female) were 
administered bentiromide as above althougr: separate urine samples were 
collected by complete bladder emptying 30 minutes, 1 , 2, 3, 4, 5 and 6 
hours after receiving the dose. This was done to provide PABA 
metabolite pharmacokinetic data. Samples were analyzed by HPLC. An 
extensive metabolic pharmacokinetic study was carried out on one of 
these subjects in which urine samples were collected at 15 minute 
intervals for the first 3 hours, 30 minutes intervals between 3 and 5 
hours and hour intervals between 5 and 8 hours post dose. 

Bentiromide was administered similarly to 10 normal volunteers, 10 
patients with pancreatic insufficiency and 5 patients with various liver 
disorders to evaluate urinary PABA metabolite concentration patterns. 
In these patients, urine was collected for both a 0-3 and a 3-6 hour 



64 



period, pooled and aliquoted so that potential differences in metabolism 
between these populations could be more readily observed. 

Colorimetric Analysis of p-Aminobenzoic Acid in Urine 



A 1.0 mL aliquot of sample urine was added to 9.0 mL of 
hydrochloric acid (1,3 H) in appropriately labeled 16x150 mm screw 
capped tubes for acid hydrolysis. The tubes were mixed and the caps 
were replaced loosely. All tubes were then placed in a vigorously 
boiling water bath and the bath was allowed to return to a vigorous 
boil. The screw caps were tightened and the hydrolysis was timed for 1 
hour after tightening the cap. After 1 hour, the tubes were removed and 
allowed to cool at room temperature for about 10 minutes. The 
hydrolysate mixture was then diluted according to the protocol outlined 
in Table 2-1 so that PA3A concentrations would be within the range of 
the standard curve. The amounts of hydrolysate and water added were 
chosen to maintain a total volume of 5 mL and the total dilution factor 



Table 2-1. Dilution protocol for urine hydrolysate mixtures. 

Urine Collection Hydrolysate Added Total Dilution 

Volume Aliquot Water Factor (D) 

mL yL mL 



Sample Preparation and Color Development 



100-200 
201-500 
501-1500 



100 

200 
500 
1CG0 



4.9 
4.8 
4.5 
4.0 



500 
250 
100 

50 



>1500 



65 



(D) accounts for the 10-fold dilution involved in the hydrolysis step. 
Sodium nitrite (0.5 mL of a 1 g/L solution in 1.3 M HCl) was then added 
to each of the 5 mL hydrolysate dilutions and to 5 mL of aqueous PABA 
standards (0, 0.5, 1.0, 1.5, 2.0 and 2.5 mg/L). These solutions were 
mixed and allowed to stand at room temperature for exactly 4 minutes at 
which time 0.5 mL of ammonium sulfamate (5 g/L) was added. The tubes 
were mixed again and allowed to stand for 4 minutes at which time 0.5 mL 
of a solution containing 1 g/L of KEDA (Bratton-Marshall reagent) was 
added. Color development proceeded for 10 minutes after mixing the 
final solution and the absorbance was determined at 550 nm against a 
distilled water reference. 

The absorbance values of standard solutions were plotted against 
their concentrations and the concentration of PABA in the urine 
hydrolysates were determined from this standard curve. The percent of 
the bentiromide dose recovered as PABA can be calculated for comparison 
to normal values by the following expression: 

% Recovered = 2-95 [PABA _ mg/L] (D) ( V) ( 1 ) 

1 bentiromide dose mg) 

where D is the urine hydrolysate dilution factor and V is the total 
urine collection volume in mL. The constant of 2.95 is derived from the 
inverse of the fraction of the molecular weight of bentiromide which is 
contributed by PABA. This factor converts measured PA3A back to an 
equivalent amount of bentiromide and also includes factors for 
conversion of yg to mg and fraction to percent. 

A hydrolysis control consisting of 320 mg/L PA ABA was taken through 
the entire analytical procedure each time assays were carried out. This 



66 



concentration of PAABA is such that 250 mg/L of PABA would be liberated 

if hydrolysis were complete and results were considered acceptable if 

PABA recovery was between 238 and 250 mg/L. This corresponds to a range 

of 95 to 100$ recovery of the hydrolysis control which was recommended 

by Adria Laboratories (144.). 

Determination of p-Aminobenzoic Acid By 
Room Temperature Phosphorimetry 

The RTP procedure for the analysis of PABA was developed first by 
the selection of the proper support material and then optimization of 
conditions for aqueous PABA standards. Steps were then taken to adapt 
the method to the analysis of urine samples from patients undergoing the 
bentiromide test and the final analytical procedure was evaluated with 
respect to various analytical variables. 
Choice of Substrate 

Two substrates (DE-61 and LTPA impregnated SaS 903 filter papers) 
were evaluated for the urinary determination of PABA by RTP. The two 
substrates were compared with respect to drying characteristics, pH 
variations, heavy atom effects and various other analytical variables. 

The drying characteristics of the two substrates were compared by 
measuring the phosphorescence intensity of aqueous PABA solutions (25 
mg/L) versus time under dry nitrogen flow. A 3 UL sample of the PABA 
solution was spotted onto each substrate (C.25 in. diameter discs) 
contained in the specially constructed sample bar. The sample bar was 
placed in the sample compartment purged with dry nitrogen. The strip 
chart recorder set on a speed of 1 in./min was started immediately after 
the sample bar was inserted. The relative phosphorescence intensity at 
420 nm with excitation at 295 nm was recorded versus time for 50 min and 



67 



each substrate was tested in duplicate. Drying characteristics on both 
substrates were also studied for PABA solutions containing 50% ethanol 
in the same manner. 

The phosphorescence intensity dependence on pK was studied for each 
substrate by adding small amounts of H2SO4 and saturated NaOK to aqueous 
PAEA solutions (25 mg/L). The H 2 S0 4 and IlaOH were added by dipping a 
small glass rod into the acid or base and transferring the adhering acid 
or base solution to approximately 3 mL of the PABA solution. The pH was 
measured by the pH meter and more acid or base was added as needed to 
achieve the appropriate pH. After pH adjustment, 3 yL of the PA3A 
solution was spotted onto the filter paper disc, samples were allowed to 
dry for 15 min in the nitrogen purged sample compartment and 
phosphorescence intensities were measured at the appropriate 
wavelengths. The DE-81 substrate was studied over a pH range of 1.6 to 
9.6 and the DTPA impregnated S<2S 903 over a range of 1.4 to 12.2. Each 
study was carried out twice and the results from all studies were 
plotted on the same graph. 

The effect of I" as a heavy atom perturber was studied for each 
substrate to determine relative phosphorescence enhancement from both 
blank and standard solutions. The blank consisted of pooled urine (pH 
6.4) collected from 6 fasting volunteers under bentiromide test 
conditions and the same urine was spiked to yield 25 mg/L of PABA to 
provide a standard solution. The RTP intensity of each solution was 
measured 10 times on each substrate with and without addition of the 
heavy atom. The heavy atom was added by spotting 2 pL of 1.0 H aqueous 
KI onto the substrate disc prior to spotting 3 uL of the test mixture. 
The results were compared for significance by the students t test (149). 



6b 



Randomly collected urine samples from 13 nonfasting subjects were 
also tested for the heavy atom effect on both substrates so that 
potential interferences from exogenous components in urine could be 
evaluated. This test was carried out similarly except that only one set 
of measurements was made for each sample and spiked samples were not 
evaluated. The results were compared for significance of the heavy atom 
effect on both substrates by a paired t test which is appropriate for 
comparison within individual samples ( 1 50) . 

The linearity, recovery and limit of detection were also evaluated 
for each substrate. The linearities of the standard curves measured 
from each substrate were determined by spotting 3 uL of each phosphate 
buffered (0.01 K, pH 6.4) standard solution, covering the concentration 
range of 5 to 500 mg/L. The recovery was determined by taking the ratio 
of the slopes of calibration curves for buffered aqueous standards, that 
were within the linear range, to the corresponding slope in pooled blank 
urine at the same concentrations. The limits of detection were 
determined for each substrate with and without the addition of heavy 
atom to buffered aqueous standards. These were calculated by 
determining the noise level and using the appropriate calibration curve 
slope to calculate the concentration that would result from three times 
this signal. 

Analysis of p-Aninobenzoic Acid in Urine 

Urine samples with collection volumes of less than 500 mL were 
diluted with an equal volume of distilled water prior to analysis so 
that the concentrations were within the range of the standard curve. A 
C.5 mL aliquot of urine or 2- fold dilution thereof from patients 
undergoing the bentiromide test was added along with aqueous standards 



b9 



(400, 300, 200, 100, and 50 mg/L) of PAEA to 0.5 mL of 1 H NaOH in 
10x150 mm screw-capped tubes, calibrated to 5 mL. The cap was replaced 
loosely and all tubes were mixed and placed in the dry heating block 
(which was pre-equilibrated to 120°C) for 1 h. The tubes were removed 
from the block and allowed to cool at room temperature for 2-3 
minutes. Four milliliters of approximately 1.0 K H 2 S0 4 containing 1.0 M 
KH 2 P0 4 was then added to each tube and the tubes were mixed. This 
mixture is a combination reagent for neutralization and buffering and 
must be titrated beforehand with the 8 H NaOH solution. The H 2 S0 4 
concentration must be adjusted according to the titration to attain a pH 
of 6.4 upon addition to the sample mixture containing 0.5 mL of 3 H NaOH 
and is done to provide an appropriate pH for RTP analysis. After addi- 
tion of the neutralization- buffer reagent, the volume is adjusted to 5 
mL with distilled water by diluting to the calibration mark on the 
tube. This allows for evaporation that may have occurred during the 
incubation step. After volume adjustment and mixing, 3 uL of this 
mixture is spotted onto a DE-81 filter paper disc which has been mounted 
on the sample bar and pretreated with 2 uL of 1.0 H KI. The sample bar 
is then placed in the phosphorimeter sample compartment and allowed to 
dry under dry nitrogen flow for 15 min. The relative phosphorescence 
intensity was measured from each position on the sample bar with the 
excitation and emission monochromaters set at 295 and 428 nm, respec- 
tively. The PA3A concentration in urine samples was evaluated by 
comparison to a best fit standard curve derived from the phosphorescence 
intensities and concentrations of the standards. The percent bentiro- 
nide recovered can be calculated by the same method used for the 



70 



colorimetric analysis and the dilution factor (D) is either 1 or 2 
depending on the collection volume. 

Ion-Pair High Performance Liquid Chromatography 
of Bentiromide Metabolites 

Urine samples from patients undergoing the bentiromide test were 
centrifuged and the supernatant (100 uL) was mixed with 400 vL of a 
solution containing 0.66 mg/L glucuronidase (156 "Fishman" units/mL) and 
0.01 H KH 2 P0 4 (pH 6.S). This mixture was incubated at 37°C in a dry 
heating block for 30 min. Following incubation, 500 uL of methanol was 
added to terminate the enzymatic reaction. These mixtures and untreated 
urine from timed specimens involved in pharmacokinetic studies were 
diluted with deionized water according to the volume voided. The 
dilution of pooled samples (0-6 hr collection period) that were treated 
with glucuronidase and samples from individual timed specimens that were 
untreated had to be carried out differently because of the 
concentrations involved. The corresponding dilution schemes are shown 
in Tables 2-2 and 2.3. The diluted urine samples were injected without 
further purification into the liquid chromatograph under the chromato- 
graphic conditions and instrumental settings outlined in Table 2-4. 
Aqueous standards containing PABA (0.12-1.0 mg/L), PAHA (0.13-1.42 
mg/L), PAABA (1.63-13-04 mg/L) and PAAHA (3.23-25.86 mg/L) were injected 
so that no more than three sample injections were carried out before one 
of the standards was injected. The peak heights corresponding to each 
analyte were measured and the concentrations in samples were obtained by 
comparison to a best fit curve constructed from the peak heights and 
concentrations of the standards. Sample concentrations were then 
corrected for dilution and volume of urine voided as shown in the 



71 



Table 2-2. Dilution protocol for HPLC analysis of timed urine samples. 



Urine Collection Added Added Dilution 

Volume Urine Water Factor 

mL UL mL 

<50 100 9.9 100 

50-100 100 4.9 50 

101-150 100 1.9 20 

>150 100 0.9 10 



Table 2-3. Dilution protocol for HPLC analysis of pooled urine samples 
following treatment with glucuronidase. 



Urine Collection Added Added Dilution 

Volume Treatment Mixture Water Factor 

mL _U]L mL 

<100 1000 9.0 100 

100-200 1C00 4.0 50 

201-500 1000 1.0 20 

>500 1000 0.0 10 



72 



Table 2-4. Chromatographic conditions. 



Parameters 
Column 

Mobile Phase 

Wavelength 

Flow Rate 

Temperature 

Chromatography time 

a.u. f .s. 

sample volume 



Conditions 
uBondapak C^o 

0.1 K TBA - methanol (90: 10) 

254 nm 

1 . 4 mL/min 

40°C 

18 min 

0.01 

20 UL 



following: 

Amount excreted (mg) = 1 COO (concentration, Ug/mL) (D) ( V) (27) 

where D is the dilution factor and V the collection volume in mL. The 
final chromatographic conditions were achieved by investigating the 
effects of mobile phase pH, the types and concentration of organic 
modifiers , buffer concentration, types and concentrations of ion- 
pairing agents, mobile phase flow rate and column temperature on the 
resolution of PAPA and its metabolites. The influence of pH on analyte 
retention was studied with a mobile phase containing W% methanol and 
0.01 I' KH 2 P0 4 . The dependence of retention on pH was also studied 
separately in a mobile phase containing 0.01 M TBA along with the other 
components. The pH of the aqueous component of the mobile phase was 
adjusted by addition of concentrated phosphoric acid or saturated sodium 



73 



hydroxide through the range of 2.6-6.0. The individual retention times 
(t R ) were measured relative to the solvent front (t Q ) and the capacity 
factors (k') were calculated at each pH and for each analyte by the 
following expression: 

k' = *R - *Q (28). 

The resolution (R g ) between each analyte at a pH selected for optimum 
separation was also calculated: 

b , (t Rl " t R2 ) (29). 

3 ~ 1 / 2 (t w1 - t w2 ) 

The quantities t R1 and t R2 refer to the retention times of the two peaks 
in question and t v1 and t V2 refer to their corrsponding peak widths 
which were measured at the baseline. 

The relative influence on analyte retention for methanol and 
acetonitrile as mobile phase organic modifiers was studied by addition 
of various amounts of methanol (20, 15, 12.5, 10, 8.7, 7.5 and 5 mL) and 
acetonitrile (15, 12.5, 10, 8.5, 7.5 and 5 mL) to a 100 mL volumetric 
flask and diluting to volume with 0.01 M KH 2 P0 4 adjusted to pH 4.0. The 
influence of methanol only (20, 15 and 1 mL added to the mobile phase) 
was also studied in the presence of 0.01 H TBA by addition of 1 mL of 
0. 1 M TBA to the flask prior to dilution. The pH of the aqueous phase 
for this study was adjusted to 6.0. Chromatography of the compounds of 
interest was carried out by injection of a standard solution for each of 
these mobile phases and the separation parameters k' and R s were 
calculated . 



74 



The effect of buffer concentration on analyte retention in an ion- 
pair method was studied by adding various concentrations (0.1, 0.05, 
0.01, 0.005, 0.003 and 0.001 M) of KH 2 P0 4 (pK 6.0) to volume in a 100 mL 
volumetric flask containing 10 mL of methanol and 1 mL of 0.1 H TBA. 
The chromatography of relevant analytes and calculation of k* values was 
carried out as before. 

A variety of both anionic and cationic ion-pairing agents were 
studied for retention behavior in mobile phases containing 10$ 
methanol. These experiments were carried out by the addition of 1 mL 
of a 0.1 M solution of each ion-pairing agent to a 100 mL flask 
containing 10 mL of methanol and diluting to volume with 0.01 M 
KH 2 P0 4 . The pH of the KH 2 P0 4 buffer was adjusted to 6.0 for cationic 
pairing-ions and to 2.5 for the anionic ion- pairing agents. The various 
agents used are listed in the materials section. Relative retention was 
evaluated by calculation of the k' parameter for analytes 
chromatographed in each mobile phase. 

Pairing-ion concentration effects on analyte retention were studied 
for TBA by adding 1, 2.5, 5, 7.5 and 10 mL of 0.1 H TBA to a 100 mL 
flask containing 10 mL of methanol and diluting to volume with 0.01 M 
KH 2 P0 4 (pH 6.0). Chromatography of the analytes was carried out for 
each dilution of TBA and separation parameters were calculated. 

The column temperature and mobile phase flow rate were varied in an 
effort to minimize analysis time. The mobile phase for these studies 
consisted of 90/10 methanol-water with 0.01 H TBA. The column 
temperature was varied by adjustment of the circulating water bath 
thermostat to 29, 40, 50, 60, and 70°C and allowing 20 minutes for 
column equilibration. A standard solution of analytes was injected 



75 



after equilibration and retention parameters were calculated. The flow 

rate of the mobile phase was adjusted to 1.0, 1.2, 1.4, 1.6, 1.8 and 2.0 

mL/min with the column temperatures set at 40°C and standards were 

injected after a 15 minute equilibration. The separation parameter k" 

and R g were again calculated following recovery of analyte peaks from 

the chroraatograph. 

Analysis of Bentiromide Metabolites by 
Room Temperature Phosphorimetry 

Separate aliquots of urine samples from clinical subjects were 
hydrolyzed by both acid and base hydrolysis. The base hydrolysis was 
carried out by the same procedure as that used for the analysis of PAEA 
in urine. Acid hydrolysis was carried out in screw capped tubes 
calibrated to 2 mL by adding 0.1 mL of the urine aliquot along with 
aqueous standard solutions containing 400, 300, 200, 100 and 50 mg/L of 
PABA to 0.9 mL of 1.3 M HC1. The caps were replaced and all tubes were 
placed in a boiling water bath for 15 min. The tubes were removed from 
the water bath and allowed to cool at room temperature for 2-3 
minutes. After cooling, 0.65 mL of approximately 2 H NaOH (titrated to 
yield a pH of 6.4 upon addition to the mixture) containing 1.0 M KH^PO^ 
was added to each tube and the tubes were mixed. The volume of each 
tube was adjusted to 2 mL, mixed by vortex, and let stand at room 
temperature until phosphorescence was measured. 

The phosphorescence at room temperature was measured from both acid 
and base hydrolysis mixtures in the same manner as for PABA analysis 
except that the substrate was not treated with iodide. The 
concentration of liberated PABA was determined by comparison to the 
appropriate best fit standard curve. The amounts of PAABA and PAABA + 



76 



PAAHA excreted were evaluated as the amounts of PABA liberated from the 
acid and base hydrolysis procedures, respectively, after correction for 
dilution and urine collection volume. 



CHAPTER 3 
RESULTS 

Development and Optimization of the 
Room Temperature Phosphorimetric Method 

The initial phase of this project involved evaluation of two paper 
substrates and optimization of conditions for room temperature 
phosphorimetric analysis of p-aminobenzoic acid. At various points in 
this study a particular condition was determined to be optimal and other 
possibilities were eliminated from further study. After the analysis 
conditions were established, a hydrolysis procedure was implemented to 
convert PABA metabolites back to the parent compound and the method was 
evaluated with clinical samples. The Bratton-Marshall colorimetric 
method was also evaluated and compared to the RTP procedure. 

The second phase of the study consisted of the development of an 
ion-pair high performance liquid chromatography method for PABA and its 
metabolites and their quantification in various populations. Again, 
analytical conditions and clinical results that were determined to be 
unsuitable were eliminated from further study. Appropriate urine col- 
lection intervals and metabolite concentration patterns were established 
during this study so that false positive bentiromide test results due to 
liver dysfunction could be detected. Finally, an RTP method for the 
analysis of the metabolites PAABA and PAAHA was developed for the 
detection of false positives due to liver dysfunction based on liquid 
chromatography results. 



77 



78 



The various analytical factors involved and the rationale for 
selecting particular analytical conditions will be briefly discussed in 
this section. A more detailed discussion of the underlying concepts 
involved in attaining these results will be discussed in Chapter 4. 
Comparison of Two Substrates 

Relatively high phosphorescence intensities of PABA have been 
observed from the paper substrates DE-81 (151) and DTPA treated S<SS 903 
(142). Therefore, these supports were chosen as potential substrates 
for the present analysis. Room temperature phosphorescence spectra of 
PABA on both substrates demonstrated excitation and emission maxima at 
295 and 432 nm, respectively, and the mono chroma tors were set at these 
wavelengths for analytical determinations. 
Drying characteristics 

The drying histogram for both substrates spotted from buffered 
aqueous PABA solutions is presented in Figure 3-1. The histogram 
consisted of 3 regions, the region corresponding to the time required 
for the RTP signal to reach a maximum (rise time), the plateau region in 
which the signal remained constant, and the decay region where the 
phosphorescence decreased in zero order fashion. The rise time measured 
from the S&S 903 substrate ( 1 3-0 min) was slightly shorter that from the 
DE-81 substrate ( 1 4- 4 min). The plateau times for DE-81 and S&S 903 
were 9.4 and 17.0 min, respectively, and the decay rate was 1.8 RPI/min 
for DE-81 and 1.6 RPI/min for S&S 903. The addition of ethanol to the 
sample solution (Figure 3-2) resulted in a slight decrease in rise time 
(14.4 to 14.0 for DE-81 and 13.0 to 10.6 for S&S 903) and an increase in 
plateau time (9.4 to 13-4 for DE-81 and 17.0-18.0 for S<SS 903) for both 



7 Q 



1/5 



0) 








3 




+J 








Eh 




© 








E 




O 








s 




o 




o 
















+^ 


n 










G 


re 








+j 


E 


03 




,0 




3 




tn 


O 






<h 


co 






a. 


.c 


a 




ft 








c 


■H 


2 


>! 


■»-» 




C 















1 





I d u 



80 



-a 


•H 





CO 




O 




05 


I 


CO 


o 


•< 


LU 


CO 




■o 


o 


o 


JC 




CO 




o 


= 






< 


co 


OL 


UJ 


1- 


Q 


Q 



I 

o 

LU 



o 








*H 




Q 




eg 








CJ 




Q 




Q 








■H 














CJ 




■* o 


<H 


Q 


o 


1 




fcn 


rj 


C 


rj 


"* 


W 


rH 


C' 






cd 










Q 


n 




rj 


cv 




CO 


03 


Q 


r] 


— 


Q 




■H 




•f- 3 


CD 


— 1 








Q 




CO 


— 






CO 




p 


ft 


o 


— 


cd 


CD 
















o 


£ 


o 


o 








■h 


GJ 




-C 


w 


-P 


c 




-p 


?h 


R) 


o 


Eh 




-*-> 




us 


E 




ED 


3 


Ph 


co 


t£ 




o 


Ph 


+-> 


dJ 


CO 


P, 


■H 


CD 


~ 


& 







C 




•H 




>> 






O 






• 





id a 



81 



substrates. The decay rate, however, was increased to 2.5 RPI/min for 
DE-81 and decreased to 1.3 RPI/min for S&S 903. 

The rise and plateau times are important criteria since they 
represent the waiting time before measurements can be made and the time 
allowed to complete the measurements. The S&S 903 substrate is superior 
in this regard since short rise times and long plateau times are 
desirable. Although ethanol improved drying characteristics somewhat, a 
dilution of urine samples was required, resulting in an additional 
procedural step and higher detection limits for the sample. It was 
decided, therefore, to use aqueous sample solutions and a drying time of 
15 min was chosen for further study of both substrates. 
pH dependence 

Figure 3-3 shows that the RTP signal of PABA from DTPA impregnated 
S&S 903 falls sharply below pH 3.3 and above pH 6.6. The DE-81 
substrate, however, demonstrated a much wider range of relatively stable 
RTP intensities between pH 2.3 and 11.7. The pH of urine can vary 
between 4.0 and 9-0 (152) so it is important to have stable RTP 
intensities over this range if urine is to be analyzed without 
treatment. This may be necessary for metabolite analysis because of the 
concentrations present, and the DE-81 substrate is superior in this 
regard. Solutions of PABA were buffered at pH 6.4 for further study 
based on these results to match the pH of the pooled urine used in some 
of these studies. 

Phosphor escence intensity and the effect of heavy atom 

Iodide has been shown to be optimal for enhancement of PABA 
phosphorescence intensities ( 1 51 > and was therefore chosen for this 
comparative evaluation. The RTP intensities of pooled blank urine (pH 



82 



33 
P 



9 
i 



6 
i 

t 



•<T-- 



X 

a. 



O 0) 

c 

T2 -H 



o 

to, 
C 

CD 

o 

a 



try 
o 
en 



i °a 

ft co 

<M T3 
O CD 

■h ta 

03 CD 
C fc. 
0! P, 

e 

C -H 





« 


cd 


PL, 


o 


Eh 


c 


n 


c 




o 




c; 




CD 


0) 


Ph 


G 


o 


■H 




H 


& 




ca 


T3 


c 


■H 




H 


p 


O 




03 


CD 




(4 


II 










CD 


.in 






o 


w 


p. (=1 


6 




a 




+3 


03 




CD 


Es 


+J 


O 


CO 


c 


U 




4J 




03 


CD 


P 




3 


-p 


03 


Ch 




c 


O 




P 


CD 


ca 


o 


p 


c 




CD 





TJ 


ss 


C 


4-> 


CD 




ft 


Ih 


CD 


o 







I 



•H 

fin 



83 



6.4) from fasting subjects and the same urine containing 25 mg/L PABA 
were compared for both substrates with and without the addition of heavy- 
atom. The results of this study are presented in Table 3-1. The DTPA 
treated S&S 903 substrate demonstrated significantly greater RTP signals 
than DE-81 in both blank (P<0.0015) and spiked samples (P<0.0025) 
without the addition of heavy atom. The DE-81 substrate, however, 
showed a significantly greater heavy atom effect in both blank 
(P<0.0025) and spiked urines (P<0.0025). With the increased intensity 
of RTP provided by the heavy atom, DE-81 proved to be the substrate of 
choice in terms of potential detection limits. The higher background 
signal observed for the DE-81 substrate is of little consequence since 
these signals are only 0.8$ of the analyte signal. 



Table 3-1. Comparison of phosphorescence intensities on DE-81 and DTPA 
impregnated S&S 903 substrates with and without heavy atom.* 



Blank Urine Urine with 25 mg/L PABA 

No I" 1.0 K I" No I" 1.0 H f 



Substrate n Mean S.D. Mean S.D. Mean S.D. Mean S.D. 

RPI RPI RPI RPI 

DE ~ 81 10 1.1 0.2 3-3 0.6 65.7 4.4 429.8 8.0 

S&S 903 

(DTPA) 10 1.3 0.4 1.7 C.4 83.5 7.7 409.3 16.0 



* n - number of determinations; RPI = relative phosphorescence 
intensities; S.D. = standard deviation 



84 



Blank analysis of randomly collected urine samples 

Potential interference from substances that may be present in urine 
from non-fasting subjects was evaluated and the results are shown in 
Table 3-2. Each sample was evaluated individually by a paired t-test 
which showed that RTP intensities without iodide added were signifi- 
cantly higher for DTPA impregnated S&S 903. The DE-81 substrate, 
however, demonstrated higher background when the heavy atom was added 
which follows the trend established for pooled urine from fasting 
subjects. The background in urine from non- fasting subjects was 
somewhat higher for both substrates with and without heavy atom which 
suggests that some interference from endogenous components of urine was 
present. The background levels with heavy atom added were still only 
0.8% of the analyte signal for DTPA treated S&S 903 and 1 .6% for DE- 
81. This indicates good selectivity of both substrates for PABA with 
regard to endogenous components of urine although DTPA treated SSS 903 
was superior. 



Table 3-2. Background analysis of randomly collected urine samples from 
non-fasting subjects (n = 13) 



DE-81 S&S 903 (DTPA) Paired "t" test* 

No r 1-3 0.3 1.5 0.3 (P<0.005) 

1 -° M I_ 7.0 4.2 3.4 1.1 (P<0.005) 



P ~ probability that the difference between reans occurred by chan 



ce 



85 



Linearity, recovery and detection limits 

The standard curve measured from both substrates was linear over 
the range of 0-40 mg/L for buffered aqueous standards. The ratio of the 
slope of the calibration curve for PABA spotted from buffered aqueous 
solutions, to that spotted from urinary solution was 0.93 for DTPA 
impregnated S&S 903 and 0.96 for DE-81. The higher recovery from urine 
observed with the DE-81 substrate could be due to the wider range of pH 
values over which the phosphorescence signal is stable. Although both 
urinary and aqueous solutions were at pH 6.4, the pooled urine has a 
different buffer capacity than the aqueous buffer solution and the ionic 
character of urinary analytes may be more easily altered in the 
substrate environment. The observation that both substrates 
demonstrated less phosphorescence spotted from urinary solutions 
indicates there may be some competition of urinary components with PABA 
for either substrate binding sites or complexation with the heavy 
atom. The limits of detection from spotted aqueous standard solutions 
was 0.26 mg/L with I" and 1.70 mg/L without I" for the DE-81 
substrate. The DTPA impregnated S&S 903 demonstrated detection limits 
of 0.26 mg/L and 1.30 mg/L with and without I", respectively. 
Optimization of room temperature phosphorimetry in urine samples 

The choice of the better substrate was based on the preceding 
studies. These studies indicated that both types of filter paper would 
be adequate substrates for urinary PABA analysis. The DE-81 substrate 
is superior in terms of pH dependence and detection limits if iodide is 
used. Although DE-81 shows significantly higher background in the 
presence of heavy atom, these levels are a small fraction of the analyte 
signal and would not cause interpretive interference. The DTPA 



86 



impregnated S&S 903 substrate had somewhat better drying characteristics 
and was superior for measurements made without iodide. An additional 
procedural step is required for DTPA impregnated S&S 903 since it must 
be prepared whereas DE-81 is commercially available. Considering all of 
these factors, DE-81 was chosen as the substrate for analysis of urine 
samples and is the only substrate used in further studies. 

Although relative phosphoresence intensity from the DE-81 substrate 
was constant through the pH range 2.3-11.7, the optimum pH was chosen to 
be 6.4. This was done to approximately match the pH found in untreated 
urine and also approach the midpoint of the pH range in which 
phosphoresence intensities were relatively constant. 

Having used iodide in the initial studies of heavy atom effects, 
various heavy atom species were investigated with the DE-81 substrate in 
an effort to optimize this effect. Thallium I, silver, and iodide were 
all compared as heavy atom species and the results are presented in 
Table 3-3. All relative phosphorescence intensities were normalized to 
the average of that found in water. The iodide perturber proved 



Table 3-3. 


Various heavy 


atom 


effects on 


room temperature 




phosphorescence int 


ensities of 


PABA from 


the DE-81 




substrate, (n 


= 5). 








Solvent 


X exc* 


Xp* 


Mean 




formalized 


(nm) 


(nm) 


RPI 


SD 


RPI 


water 


294 


423 


38 


1.0 


1 .00 


0.1 M KI 


294 


428 


120 


8.9 


3.16 


0.1 H TINO^ 


294 


428 


22 


1.4 


0.58 


0.1 H AgKOj 


294 


428 


21 


1.7 


0.55 


* ^exc and A 


p = the wavelengths 


of maximum 


excitation 


and emission 


respectively. 











87 



superior by increasing phosphorescence intensities 3.16 fold. The 
phosphorescence intensities in the presence of the two positively 
charged species (Tl + and Ag + ) were decreased by about 50% in each 
case. This is consistent with results observed for Ag + by Su and 
Winefordner ( 1 5 1 ) but inconsistent with their results for Tl + . A 
possible explanation for this may be that a different counter-ion 
(chloride) was used for thallium in the previous study. The reduced 
phosphorescence intensities that were observed in this work may be due 
to competition of the N0^~ anion with analyte molecules for adsorption 
sites on the substrate. Iodide was chosen for further study because of 
the higher relative intensities which provided better potential for low 
level detection. 

The optimum hydrolysis conditions (to convert PABA metabolites back 
to the parent compound) were determined by comparing acidic and basic 
hydrolysis procedures and analysis by the HPLC procedure. The stability 
of PABA to hydrolytic conditions and the completeness of hydrolysis of 
the metabolites were used as selection criteria. Figure 3-4 and 5-5 
show the time course of acid and base hydrolysis of metabolites respec- 
tively. The acid hydrolysis does not completely convert p-acetamido- 
hippuric acid (a major PABA metabolite) to the parent compound as shown 
in Figure 3-5. The basic hydrolysis procedure, however, demonstrated 
complete hydrolysis of all metabolites to PABA which was completed in 
approximately 1 hour. The parent compound was stable to conditions of 
hydrolysis for both metabolites as indicated by a relatively constant 
concentration profile after the plateau region had been reached for both 
experiments. 



88 



100 h 




-o- 



-□ 



Fig. 3-4. 



20 40 

T! ME (Minutes) 



60 



80 



Percent of p-aminobenzoic acid recovered versus time from p- 
acetarcidobenzoic acid (open squares) and p-acetamidohippuric 
acid (open circles) under acid hydrolysis conditions. 



89 






Fig. 3-5 



20 40 

TIME (Minutes) 

Percent of p-aminobenzoic acid recovered versus tine fron p- 
acetanidobenzoic acid (open squares) and p-acetar.idohippuric 
acid (open circles) under alkaline hydrolysis conditions. 



90 



At this point, it was decided to investigate why the acid hydro- 
lysis procedure had been used successfully in colorimetric methods and 
experiments were carried out to explain why low recovery of PABA meta- 
bolites had not been observed previously. Para-aminohippuric acid (the 
primary compound to which PAAHA is converted during acid hydrolysis) was 
taken through the Bratton-Karshall colorimetric procedure along with an 
equimolar solution of PABA. The acid hydrolysis step was deleted from 
the procedure and the chromophores were scanned by absorbance spectro- 
scopy (Figure 3-6). Figure 3-6 shows that PAHA apparently reacts to 
form a chromophore with spectroscopic properties very similar to that of 
the PABA-chromogen complex and although conversion to PABA was not com- 
plete, this could not be detected by the colorimetric method. The RTP 
method, however, is more selective for PABA and the acid hydrolysis 
procedure would be inadequate because of low recovery of PABA metabo- 
lites and variable PABA metabolism. 

This was further investigated by recovery experiments in which PABA 
and equimolar solutions of PAKA, PAABA and PAAHA were taken through the 



Table 3-4. Recovery of PABA and metabolites by acid hydrolysis followed 
by colorimetric quantification (n = 5) 



PABA 
PAHA 
PAABA 
PAAHA 



Mean concentration 
(mg/L) SD 

2.48 0.03 

2.44 0.03 

2.34 0.04 

2.29 0.02 



Mean percent 
recovered 

99.2 

97.6 

93.6 

91 .7 



91 



< < 

n z 

< < 
o. a. 




E 

c. 

tr» , - 



- g 





H 




CC 


CO 






i — i 




o 


o 




»H 


E 


* 









t-3 


(u 




!h 


5 


o 




cQ 






DQ 




C 




CD 




3) 


•H 






Ph 






£h 






£3 


















CQ 


ft | 




CO 


c 


_ 
Eh 


-T 




CD 








CD 






H 


c 






■H 


M- 








CD 


u 


u 


CD 











CD 


£h 


C 


►> 

1* 


ft 








Ch 


CO 


H 




X3 


i — 1 


G 


Eh 


cu 


C 


— 




■ H 





CQ 


~* — 


Pi 


Li 


O 


6 






c 




o 


CJ 


1 


M 




C 






c 


I — 1 






1 — 1 


o 














E 


DQ 




pq 


t. 


• 




CO 




CD 












■*-> 


Eh 


< 




c 


— 


CD 






Ih 






O 


CC 




r ~ 


Ph 


■H 


CD 


PQ 


CJ 






£3 




QJ 






-C 


u 






•h 




Sh 




Ph 




o 


o 


ft 






ft 


X 




■H 


C\j 


E 




SO 


3 




• 


tn 


c 




-(-> 


•H ■ 




u 


E 




a 


ca 


CQ 




i 


C 


0} 


ft 


c 






H 


c 


T3 


■P 


o 


c 


CO 


H 


EC 





& ^ c 

U < CD 

O sq o 

W < c 

— ft o 



o 
i 



92 



entire colorimetric procedure which included acid hydrolysis. The 

results of this study are shown in Table 3-4 and indicate that PAHA 

demonstrated the highest recovery whereas PAABA and PAAHA were 

successively less recovered. This is probably because of incomplete 

hydrolysis of the amido groups on these compounds and the fact that the 

amides do not react with Bratton-Marshall reagent to form a chromophore. 

Clinical Evaluation of the Room Temperature 
Phosphorescence Method 

Having developed and optimized the RTP method, studies were carried 
out to evaluate the method in spiked urine and in urine samples taken 
from subjects undergoing the bentiromide test. The method was evaluated 
for recovery, precision, selectivity, and various other analytical 
variables. The method was also compared with colorimetric analysis by 
correlation studies. 
Analytical Recovery 

Para-aminobenzoic acid (250 mg/L) and its metabolites PAHA (354 
mg/L), PAABA (326 mg/L) and PAAHA (431 mg/L) were diluted in pooled 
blank urine, taken through the alkaline hydrolysis procedure, and 
evaluated by the RTP method (Table 3-5). These metabolite concentra- 
tions are such that an equivalent amount of PABA (250 mg/L) is liberated 
for each compound if hydrolysis is complete. Preliminary experiments 
demonstrated that the phosphorescence intensities produced by the 
metabolites studied were all less than 10$ of that produced by PABA 
(Figure 3-7). Therefore, any appreciable signal detected from the 
hydrolysate solutions was due to liberated PABA and not from metabolite 
phosphorescence. 



9? 








ca 












Z 






T3 




' s 


CD 






M 


1 


v — 








CO 




T3 


■V 


* 


•H 






O 


E 




n 








m 




o 


CD 




•H 




■H 


o 




O 


N 




CC 


c 






33 


' — ' 




& 






o 


Cm 




C 


o 




•h 




Pi 


e 


E 




CtJ 




■H 


1 






ft 


-m 


c 




u 


c 


T3 




•H 


c 


ft 


R 


ca 


co 


ca 






i 




0) 


ft 




c 






c 


Cm 




o 


C 


~ 


c 




■H 


CO 


Kl 


U 




m 






-M 




c 


O 


y 




CD 


■H 


ft 


ft 


m 


co 



co 



2 o 
ft a 
ft ft 



o 


•H 






•H 




cu 




w 


c 






CO 


■a 






■H 


•H 






E 


E 




• 


<y 


CO 




c 




+3 




o 


o 


11 




•H 


o 







-P 


a 


CC 


M 


CO 


a 


1 


-(-> 


o 




ft 


co 


■H 


co 






Cm 


o 






•iH 


M 




co 


(-1 


o 


CM 




ft 








E 






03 


C 


CO 








c 


•H 




f-l 




O 


Q 


ca 


c 


CC 




c 

60 


0) 


o 




•H 




•H 


+J 


K 




o 






+J 




O 


M 


.3 


C 




<D 


M 


O 


c 


3 


a 




o 


O 


ft 


o 















!) -H O 'D 
-M E ,0 rH 



. M O 

-P C Cm 
CO CO I 

OuO 
a ca t- 



id a 



94 



Table 3-5. Analytical recovery of PAEA and metabolites added to drug 
free urine by room temperature phosphorescence 



Concentration of PABA (mg/L) 





n 


Added 


Mean 


SD 


Mean {%) 


SD 


PABA 


10 


250 


253 


7.1 


101 


2.6 


PAHA 


10 


354 


265 


8.3 


106 


3.3 


PAABA 


10 


326 


245 


4.2 


98 


1 .7 


PAAHA 


10 


431 


244 


8.9 


98 


3.6 



Table 3-6. Precision of the room temperature phosphorescence method 



Within- run 



Day-to-day (mg/L) 



Mean (mg/L) 


n 


S.D. 


C. V. (%)* 


93.2 


10 


7.9 


8.5 


162.0 


10 


8.6 


5.3 


347.2 


10 


21 .8 


6.3 


Mean 


n 


S.D. 


C.V. {%) 


75.5 


10 


10.9 


14.4 


163.6 


10 


9.0 


5.5 


335.5 


10 


19.2 


5.7 


of variation. 









95 



Precision 

Precision was evaluated by repeated analysis of frozen aliquots 
taken from patient samples on both a within-day run and day-to-day basis 
(Table 3-6). The three patient samples chosen provided concentrations 
reflecting the entire range of the assay. The coefficient of variation 
(CV) for all samples was less than 10%' except that for day-to-day 
precision of the sample with the lowest concentration. 
Linearity 

The standard curve constructed from standards that were subjected 
to the alkaline hydrolysis step was linear through the range of to 40 
mg/L (Figure 3-8). The range of linearity is the same as that for 
aqueous standards that were not hydrolyzed which shows that the linear 
range is not affected by the concentrated electrolyte environment of 
hydrolysate mixtures. The linear range corresponds to original urinary 
PABA concentrations of to 400 mg/L and covers most concentrations 
found in patient samples. Of the 75 patient samples analyzed, only 6 
required dilution with an equal volume of distilled water prior to 
hydrolysis to adjust their concentrations into the range of the standard 
curve. 
Sensitivity 

The limit of detection measured with standard solutions of PABA 
diluted in pooled blank urine was 0.67 mg/L. This detection limit is 
slightly higher than that measured for iodide treated DE-S1 with aqueous 
standards (0.26 mg/L) and is due to the higher phosphorescence intensity 
of the pooled urine relative to that of water. The detection limit in 
urine, however, is far below the level found in patient samples. 



9o 



< 
m 







■H 




o 










■a 


O 


Ch 


•H 


a 





TJ 


N 


C 


a 


C3 


CD 


4* 




DQ 







a 





•H 


x: 


£ 


+j 


03 




I 


%H 


ft 


o 


in 


0) 




to 


(0 


c 


In 


CO 




Eh 


> 






Ih 




Ed 


ti 




3 


a 


4-> 


■H 




rH 


L, 









P. 




-t-> 


o 










C 


S 


H 


o 


3 


o 


O 


c 






09 








a> 



>> C6 
-P Si 
•H -P 

m cq 

C ,0 
O D 

c 

CO 

1) I 

o 55 

C ft 
<JJ 

o o 

H 
tt) 



O f-i 

.c o 

P, <M 

CO 

o c 
,c o 

P. -H 

-p 

> 

•H -t-> 

c 

CQ O 
' — f o 

co c 
o 
o 



o ^-n 

< 

-p ffi 
o < 

rH C3-, 
Oh 



I 



•H 



97 



Selectivity 

Several commonly used drugs were tested for interference with the 
RTP method by adding 500 mg/L of each drug to pooled blank urine and 
analyzing these solutions for PABA. The only drug that demonstrated an 
interference that would preclude clinical evaluation was procaine (Table 
3-7). The consequence of the procaine interference is that patients 
must be screened to see if they have recently received the drug. Most 
of the drugs tested have been shown to interfere with colorimetric 
methods (see Appendix I) and this study illustrates the superiority of 
the RTP method in this regard. Xylose was tested because of the possi- 
bility of administering the xylose test concurrently with the bentiro- 
mide test. 



Table 3-7. Drug interference study based on analysis of aqueous 
solutions of selected drugs (500 mg/L) by RTP. 



Apparent PABA 
Drug concentration (mg/L) 


Percent 
Interference 


Chlorthiazide 








Sulfadiazine 








Sulfamethazine 








Atropine 








Neomycin sulfate 








Indomethacin 


20 


4.1 


Chlorpropamide 








Tolbutamide 








Methochlopramide 








Hydrochlorthiazide 








Acetaminophen 








Procaine 


203 


40.6 


Acetyl salicylic acid 


5 


1.1 


Lidocaine 





C 


Caffeine 








Xylose 





c 


Sulf anilimide 


40 


8.0 


Chloramphenicol 


i. 
j 


0.7 



98 



Patient Sample Correlation 

Urinary PABA concentrations measured by the RTP method were com- 
pared with those obtained by the Bratton-Marshall colorimetric proce- 
dure. Urine samples were collected from patients undergoing the ben- 
tiromide test according to the protocol in the methods section of this 
work. Consequently, drug interference was not a factor. The two 
methods demonstrated good agreement as indicated by the correlation plot 
shown in Figure 3-9. Regression analysis of the Bratton-Marshall re- 
sults (x) versus the RTP results (y) demonstrated a linear relationship 
of y=0.997x+1 .651 for 75 clinical samples containing between 58 and 786 
mg/L. The correlation coefficient (r) was 0.993 and the standard devia- 
tions of the slope and intercept were 0.002 and 0.300, respectively. 

Development and Optimization of the Ion-?air High 
Performance Liquid Chromatography Method 

Ito and coworkers have quantified PABA by HPLC in urine samples 
after metabolite hydrolysis (153). The goal here, however, is to 
quantify individual PABA metabolites so that false positive bentiromide 
test results due to liver dysfunction may be detected. For this 
purpose, a new analytical approach which would resolve metabolites of 
PABA from each other and also from endogenous components of urine was 
necessary. After initial experiments, conducted with various organic 
modifiers and at different pH, it was decided to employ ion-pairing 
agents which would enhance the retention of the compounds of interest 
without affecting the endogenous components of urine. Both anionic and 
cationic ion pairing agents were tested since all analyte molecules are 
amphiprotic. The ion- pair method that was ultimately developed was then 
optimized and evaluated. 



99 



- 



s 




o 


CO 


o 


Ph 


u 


to 






cl> 






c 


■H 


o 






>> -P 




B) 




M 




EC 


9) 




Eh 


CO 




js 


m 


■p 


ea 




o 


£ 


E 




O 


•a 


c 


cu 


•H 


(l, 




3 


3 


CO 




CO 


C 







E 



o 

2 



E 1 



o 
o 



o 

o 

CN 



O 

o 



o 
o 
<o 



o 
o 



O 

o 

CN 



CO 



o 


















o 


T3 




•H 


a) 




o 


CJ 


II 


N 





c 


C 








ft 




,C 






C 


0! 




G 


O 




•H 


c 




E 


o 


• 


CO 


V 




1 


a 




ft 


CD 


+ 




u 






c 


X 


o 


JC3 






ft 


CTi 


-P 


CO 




O 


o 


• 


iH 


Si 


O 


ft 


ft 


II 






>5 




CD 




O 


f-. 




•H 


3 


X 


-P 


-P 




CO 


CO 




tH 


tn 


■o 


a> 


CD 





h 


ft 


si 




s 


-p 


O 


11 


a; 


o 


■P 


E 



(1/BLU) diy 



i 



•H 



100 



Organic Modifier 

The influence of methanol and acetonitrile concentration on the 
retention of the compounds of interest was studied without the use of 
ion-pairing agents. The aqueous component of the mobile phase was 
buffered at pH 4 with a 0.01 M phosphate buffer and the column was left 
at ambient temperature (25° C). In all cases, decreasing the 
concentration of organic modifiers increased the retention of the 
solutes by the column (Figures 3-10 and 3-11). However, acetonitrile 
had a more selective effect on retention compared to methanol. For 
example, lowering the acetonitrile concentration from 15 to 5% increased 
the capacity ratios of PAABA and PAAHA from 1.0 to 2.4 and 0.4 to 1.3, 
respectively. The corresponding increases in the capacity factors of 
PABA and PAHA were from 1.0 to 1.3 and 0.4 to 0.6, respectively. 
Consequently, 5% acetonitrile provided satisfactory overall retention 
characteristics, but resolution of PABA and PAAHA was poor (Figure 3- 
10). In comparison, methanol demonstrated a less selective effect on 
the capacity factors of PAABA and PAAHA which increased from 2.6 to 4.1 
and 0.9 to 2.3, respectively, when the methanol concentration was 
decreased from 20 to 5% (Figure 3-11). The corresponding increases in 
the capacity factors of PABA and PAHA were 1.0 to 1.4 for PABA and 0.4 
to 0.5 for PAHA. With concentrations of acetonitrile above b%, the 
solutes eluted too rapidly for analytically useful results. 

The results indicate methanol is a stronger solvent than 
acetonitrile for PAABA and PAAHA than for their non-acetylated 
counterparts. A possible explanation for this is that amides (i.e., 
PAABA and PAAHA) are generally stronger in their hydrogen bonding 
ability than amines (i.e., PABA and PAHA) ( 154) and could be expected to 



2.5£, 




Acetonitrile (%) 



The effect of mobile phase acetonitrile concentration on 
capacity factors for p-aminobenzoic acid and netabolites (p- 
aminobenzoic acid, closed circle; p-aminohippuric acid, 
open triangles; p-acetamidohippuric acid, open circles; p- 
acetanidobenzoic acid, open squares). The chromatographic" 
conditions were the same as those listed in table 2-4 except 
the flow rate was 1.2 ml/min and the column temperature was 
25°C. 



102 




Fie- 3-11. 



The effect of mobile phase methanol concentration on 
capacity factors for p-aminobenzoic acid and metabolites. 
The symbols and conditions are the same as for figure 3-10. 



103 

interact more strongly with a hydrogen bonding solvent. Methanol can 
serve as both a proton donor and acceptor (155) making it a stronger 
solvent than acetonitrile for hydrogen-bonding interactions (156). 

With a mobile phase containing 10$ methanol the resolution of all 
the peaks of interest was greater than 1.00. Although a mobile phase 
containing W% methanol in a phosphate buffer adequately resolved the 
peaks of interest, interference from endogenous components of urine was 
observed. Complete resolution was achieved by optimizing the effect of 
pH and by the addition of ion-pairing agents to the mobile phase. 
pH Optimization 

The dependence of retention on pH was studied with a mobile phase 
containing W% methanol at ambient temperature. The pH of the aqueous 
component of the mobile phase was adjusted by addition of phosphoric 
acid or sodium hydroxide to 0.01 M KH 2 P0 4 . Figure 3-12 shows that the 
retention of all components decreased upon increasing the pK above 3.5 
due to deprotonation of the carboxylic groups of the solute (157). 
Changes in the elution order with changing pH nay be attributed to 
differences in the pKs of the carboxyl groups. For example, the pKa 
values for PABA and PAHA are 4.9 and 3.6, respectively (158). No 
literature values for the pKas of PA ABA and PAAHA were available. These 
parameters were calculated from the data in Figure 3-12, however, and 
pKa values of 4.27+0.07 for PAAHA and 4.23+0.04 for PAABA were 
obtained. The pH ranges used for the pKa calculations were 3.9-4.6 for 
PAAHA and 3-9-4.7 for PAABA. The pKa of PAABA was also determined by an 
absorptiometry method as a check on the value obtained from 
chromatographic data and was found to be 4.18+0.05. The close 



1C4 



cd 
a 

ft o 




- CD 



CP 



CM 



CO 

d 



CD (D 

Eh -H 
XI 

c 
• E 

03 

Hi CD 
-P £ 
•H -P 



c 

o 

■H 
■H 

c 

•rl 
«H 
CD 
T3 



O 



SJ CD 

o 



o 
-p 



•u O 
c <- 
a i 

•H CD 
U Si 
CO 3 

O -H 
•H <M 

o 



ro 



d 





C 


M 


— 


•H 


3 










CU 


•H 


o 


03 


<h 




o 




•H 




CD 




-p 


SI 






CO 


' 7 


03 




L ft 


CO 




m 


gj 


— 


o 


E 


o 


Cm 


Q 


Oh 




03 


C\J 


03 






M 


CD 











-p 






c 








a 








o 




<d 




>s 






-p 








03 




o 


C 


c 




o 


CD 


ft 


•H 




ca 


-P 




o 


•H 


O 




T3 




c 


C 


CO 


o 


o 


— 




o 


■p 






CD 


"ft 


o 


E 




■H 




<M 






o 


ft o 




CO 




-p 


(h 




o 


ttf 




o 


o 


CD • 




+■> 


C CO 


Cm 


£3 


■H i-( 




5 


E O 




o 


•P -Q 


CD 


Ih 


C E 




.C 


c >> 




CJ 


O 03 



I 



• 

Si 
■H 



105 



comparison between the two values indicates that the pKa's measured 
chromatographically were reliable. 

Separation was achieved at pH 4.0. Under these conditions, 
however, components of human urine persisted and ion-pairing agents were 
necessary to enhance the retention of the compounds of interest without 
affecting the endogenous components of urine. 
Ion-pairing Agent 

The increase in retention provided by ion-pair formation is 
dependent on the charge, hydrophobicity and size of the ion pairing 
agent (155). Increasing the pairing-ion alkyl chain-length for a 
homologous series of symmetrical counter ions should, therefore, result 
in an increase of log k' with increasing counter ion carbon number 
(160). 

Various alkylammonium ion- pairing agents (0.01 M) were studied for 
their effect on analyte capacity factors and the results are shown in 
Figure 3-13. The positive deviation from linearity demonstrated for the 
hexadecyltrimethylammonium ion (19 carbons) is possibly due to a 
secondary equilibrium in which these surface acting pairing ions are 
adsorbed onto the surface of the stationary phase (161). This could 
result in alteration of the stationary phase by increasing its effective 
hydrophobicity and creating an apparent ion-exchange effect. Both of 
these factors would contribute to the observed positive deviation. 
Another explanation for this deviation from linearity may be that the 
charged head of the asymmetric hexadecyltrimethylammonium ion is more 
accessible for ion pairing than those of the symmetrical series 
represented by the other pairing ions. The homologous series of 
symmetrical pairing ions have the bulky alkyl groups bonded directly to 



106 




0.9 ' — 1 — 1 1 — 1 — 1 — i — ' ' i ■ 

4 8 12 16 19 

Counterion Carbon Number 



Fig. 3-13. Cationic counter-ion size effect on capacity factors for p- 
aminobenzoic acid and its metabolites. Tetramethylamnonium 
chloride = 4 carbons; tetraethylammonium chloride = 8 
carbons; tetrabutylamnonium chloride = 16 carbons and 
hexadexyltrimethylanmoniurn bromide = 19 carbons. The 
symbols and conditions were the same as for figure 3-10 and 
the pH of the mobile phase was adjusted to 6.0. 



107 



the ammonium ion which may cause steric hindrance with binding to solute 
ions. Increased availability of the charged head on the long chain 
alkylammonium ion would result in a higher ion-pair formation constant 
and thus a higher extraction constant (E). This would be expected to 
increase solute retention according to equation 25 in the introduction. 

Anionic ion-pairing agents were also studied for their effect on 
retention of the compounds of interest with the aqueous component of the 
mobile phase adjusted for pH 2.5. Figure 3-14 shows a plot of analyte 
log k" values versus counter ion carbon number for the anionic pairing 
ions studied. The increase in retention for PAHA and PAAHA is 
approximately linear whereas log k' for PAABA and PA3A show deviations 
from linearity above carbon number 8. The ion-pairing agents used 
constitute a homologous series of alkyl sodium sulfates so any 
deviations from linearity are probably not attributable to accessibility 
of the charged head on the pairing ion to a site of positive charge on 
the analyte. Also, these deviations occurred for both PAABA which is 
unionized at pH 2.5 and PABA which is partially ionized. This suggests 
that the deviations observed were not due to the formation of ion pairs 
because PAABA would not form ion pairs with the pairing agent. The most 
reasonable explanation is that the stationary phase was altered by 
absorption of pairing ions. This could result in a decrease in 
retention for PAABA and an increase in retention for PABA because the 
stationary phase would take on an anionic character which would result 
in stronger interactions with positively charged molecules and weaker 
interactions with uncharged species. This explanation is consistent 
with the positive deviations in retention observed for PABA and the 
negative deviations observed for PAABA. 



1CS 




- CM 



CD 
-Q 

e 

3 



C 

o 

-O 
i_ 
O 

o 



00 



c 

o 

c 

o 
cj 



CM 

d 
i 



i 

ro 
d 



d 



CD 



CO CO 



■H 



C 

c 

ID M 
E C3 

a o 



o 

e 
>> 

CO 

co 

Q) as 
js * 

0) 
CO 

• CO 



T3 


ac 


CO 


A 


■H 




c 


P-. 


w 


ii 


o 




CD 




& 


CO 




o 


u 


■■ — i 





+j 


CO 


■H 


•r— 


CO 


CJ 




C 






O 




i — 1 


C\J 


E 


c 










Z~. 




1) 


& 




II 




O 


E 






c 


3 


cd 




•H 


■H 


+j 


Ch 


B 


TJ 


d 


o 


c~ 


o 


C)H 






co 


1 1 


PC 












3) 


CO 






c 




a 


C 


CO 


E 




p . 


-•— 










•H 




CO 


o 


~ 




*- 




C 


c 


Q 




co 


CO 


+J 








o 


CO 


CJ 


o 






c 






o 








JO 


o 




>a 


H 








CD 








U 


c 


(h 






— 






vo 






ft 




T3 


•H 




ii 


C 


C(H 


o 




s 






CD 




H 


c 


4J 


CO 


o 




CO 


c 


Ch 




<~ 


o 




4-> 


H 


fit 


CO 


o 




'- 


CO 


CD 


CO 






Ch 




'6 


0) 




E 




E 


CD 






CO 




■H 




CO 


cd 


■a 








o 


ii 


CD 


•H 


CO 




X3 





CD 


-4-> 






C 


c 









c 


CO 


Ch 


H 




•H 




r— 1 






1 


CD 


3 




• 


~ 


— 


co 




cr> 


CD 






co 










a 




c 




3 


o 






co 


•H 


•H 


o 


O 


CD 


T3 







CJ rH 

•H O 

C XI 

O CC 
■H 

C CD 

< E 



I 



T3 

c 
o 

O CO 



CD 



o -c n 

CD C T3 

tj cc a 



•H 

fa 



109 



The dependence of analyte capacity factors on pK without pairing 
ion (Figure 3-12) indicates that both PAABA and PAAHA are unionized at 
pH 2.5 whereas PABA and PAHA are at least partially ionized. A greater 
effect on retention should be observed for the more ionized species when 
the pairing ion is increased in size and the greater ion-pair effect 
observed in Figure 3-14 for PABA and PAHA is probably a result of this. 

Octanesulf onic acid was also compared with octane sodium sulfate to 
determine the effect of a different charged head group on the pairing 
ion. The results of this study are shown in Table 3-8 for each of the 
analyte molecules. The retention of all solutes was increased by the 
substitution of the sulfonate for the sulfate suggesting a higher 
formation constant for the former pairing ion. The most likely reason 
for this observation is higher negative charge density on the sulfonate 
molecule which results in stronger association with the cationic pairing 
agent. 



Table 3-8. Comparison of octane sodium sulfate with octane sulfonate 
for retention of PABA and its metabolites 

Octane sulfate Octane sulfonate Percent 
k kj difference 

0.79 0.83 5.1 

0.73 0.80 9.6 

1.22 1.28 4.9 

0.47 0.60 27.7 



PABA 
PAHA 
PAABA 
PAAHA 



110 



The cationic pairing ions demonstrated greater increases in 
capacity factors for all analytes when compared to the anionic pairing 
agents. This, no doubt, can be attributed to the higher fraction of 
solute ions at pH 6 than at pH 2.5 (see Figure 3-12). The bonded phase 
column used is not stable below pH 2.5 so lower pH values could not be 
employed to further ionize solute molecules. Furthermore, the retention 
of PAHA was only slightly affected by anionic pairing and could not be 
resolved from fast eluting endogenous urinary components. It was 
decided, therefore, to use a cationic pairing agent for further study 
and tetrabutylammonium ion (TBA) was chosen specifically because it 
provided a good compromise between high capacity factors, needed for 
resolution from interferences, and fast analysis time. 

The effect of methanol concentration on retention of the compounds 
of interest followed the same trend with 0.01 M TBA added (Figure 3-15) 
as that shown in Figure 3-11 without TBA. Again, the best separation 
was achieved at \0% methanol. 
Pairing-ion Concentration 

The effect of TBA concentration on solute retention was studied at 
pH 6.0 and ambient temperature in a mobile phase consisting of 10 v/v>' 
methanol and 90% aqueous phosphate buffer. The addition of TBA (Figure 
3-16) to the mobile phase resulted in a reversal of relative positions 
for PABA and PAHA along with PAABA relative to PAAHA, which was due to 
the greater ion-pair effect on the glycine conjugates. Analyte capacity 
factors were studied through the range of 0.C01 to 0.01 M TBA and the 
capacity factors increased from a range of 0.11 to 0.50 at 0.001 K TBA 
to a range of 0.92 to 2.50 at 0.01 K TBA, without any significant change 
in separation selectivity. This agrees with previous studies ( 161, 162) 



1 1 1 




0' ' ■ L_ 

10 15 20 

Methanol Concentration (%) 

Fig. 3-15. The affect of mobile phase methanol concentration on 

capacity factors for p-aninobenzoic acid and its metabolites 
with tetrabutylammonium chloride (0.C1 M) added to the 
mobile phase. The symbols and conditions were the same as 
for figure 3-10. 



112 




cm « 



o 
d 



10 

O 

o 
d 



m < 

o m 

Q h 

o - 



in 

OJ 

O 
O 

d 



o 
o 
d 



o 

ro 



Q 

CO 



o 

o 

CD 

w en 
Si o 
o JZ 
4-> -(-> 

u 

a: ra 

>-. CD 

-p e 

•H CO 
O CO 

a) 

ft -i 
re j= 
o -P 

CD CD 

-t-> X 

C CO 
O C 
O 

C -H 
O -P 
■H -H 
4J -D 

ea c 
w o 

4J o 

c 

CD 

o c 

C CO 
O 

O CO 
i—f 

o 

« E 
Eh >: 

— co 

3 CD 
■H Eh 



O 



o 
-p 

T3 
CD 
4J 
CO 
3 



O 

co 
ta- 



xi co 

O CD 
-P 
E -H 

3 r-l 
■H O 

c 

o cc 

E -P 



o 

H 

CD 
+J 



S3 E '-w 



1 — 1 




O 






-P 


c 








ft 


.S 






CO 


T3 


CD 


Pn 


■H 






C 


-P 


CD 


a 




+-> 








o 


c 




■H 


a 


c 


Q 






N 


o 


-p 


a 




o 


CD 


i 


CD 


~ 


t<*\ 




O 






(3 


CD 


i 


■H 






E 


- 


CD 


ca 


feC 


£3 




•H 


Eh 


ft ^ 



c 



I 



113 



where it was seen that pairing-ion concentration had no effect on 
selectivity although solute ion retention is significantly affected. 
The relationships between TBA concentration and solute capacity factors 
are sigmoidal which has been previously observed (162). This is also 
consistent with low pairing-ion binding to the stationary phase since 
surface acting pairing-ions usually show a parabolic dependency of 
capacity factors on pairing-ion concentration ( 1 62 ) . For the purpose of 
method development, it is advantageous to identify the maximum effect 
and use the pairing-ion at this concentration, which in our case was 
0.01 H. 

Optimization of pH With the Ion-Pairing Agent Added 

The dependence of retention on pH with TBA added (Figure 3-17) was 
much less pronounced than that shown without TBA in Figure 3-12. The 
capacity factors of all compounds were increased as the pH was raised 
above pH 5.0 showing the increased ion-pairing effect with increasing 
concentrations of free carboxylate ions. Between pH 4.0 and 5.0 the 
relationship can best be described as a combination of ion-pairing and 
ion-suppression effects since the weaker acids PABA and PAABA 
demonstrated the greatest relative increase in capacity factors. The 
minimum capacity factor for all compounds occurred at about pK 5.0 
suggesting a common point at which the ion-pair effect begins to 
dominate over the ion suppression effect. This is an interesting 
observation since it seems that the greater ion suppression effect of 
PABA and PAABA is about equally compensated by greater ion-pair effects 
on PAHA and PAAHA at that pH. 

Separation was achieved with a resolution of 1.0 between PABA and 
PAHA, 1.14 between PAHA and PAABA, and 2.0 between PAABA and PAAHA at pH 




o 



•H 


I_ 






0) 










CO 








o 










*H 






! 1 


c 




Q 












to 


o 










CD 


■ 
■ 










CO 


Q) 




4-3 


;_ 




'H 












X3 












— 








nj 




ti 


(U 




•H 


~Z 






















Q 


o 


m 






i — j 


o 




o 


V" 


— 


.- j 




ft 




CD 




>-. 




CD 


w 


Q 


1 — 1 




c 


•H 





•H 






Ez 


o 




CC 


E 




1 




V-t 




<d 


c 














o 




o 




o 


.f-i 




-p 




CO 




•H 




T3 




o 








XI 




o 


■a 




















+J 


•H 




•H 


Ph 


rj 


CJ 


O 




»u 






ft ^ 


OT 


cc 


CJ 




o 




r\ 




e 


■H 


C 


s 




c 


•H 


•H 




C 


— 




O 


C 


"ft 


i 


o 




E 


o 




cfl 




O 


i — 1 


r— I 




>5 


Cfl 




*J 


■P 


CJ 






o 




o 


Cm 




E 


Sh 


Fh 


■H 










JJ 






+■> 


ft 






X 


EH 







i 



■H 



115 



5.8. Addition of TBA also increased the retention of urine contaminants 
slightly to a capacity factor of 1.1. 
Effect of Buffer Ions 

Buffer ions can compete with ion-pairing agents for analyte 
molecules, thereby, reducing the influences of the latter (153). The 
effect of phosphate concentration on analyte capacity factors is shown 
in Figure 3-13. An average decrease in analyte capacity factors of 20$ 
was observed between 0.0001 and 0.005 M phosphate without changing the 
elution order of sample components. Therefore, it was decided to 
eliminate the phosphate buffer at this point and use TBA alone for 
further study. The TBA alone provided a stable pH of 7.4 and obviated 
the buffer effect without affecting peak shape. 
Temperature and Flow Rate 

Relative separation was investigated between the temperature and 
flow rate ranges of 29-70° C and 1.0 to 2.0 mL/min, respectively. Plots 
of capacity factors versus temperature (Figure 3-19) and flow rate 
(Figure 3-20) show little variation because capacity factors are 
measured relative to the solvent front which is also affected by 
temperature and flow rate. Inspection of the individual chromatograms, 
however, allowed the optimization of variables with regard to providing 
fast analysis time while still maintaining separation with all 
resolution parameters greater than one. The optimum combination of 
temperature and flow rate chosen in this manner was 40° C and 1.4 
mL/min. At this point, all compounds of interest had been adequately 
separated and resolved from endogenous components of urine. Example 
chromatograms from blank urine and urine spiked with the compounds of 



116 



o □ <& 



o □ <© 



o a <&■ 




i 

Q 

C\J 



o 
o 



— ii 

o o 
d cl 



in 
O 

o 
o 



CO 

•H 

0) 

T3 r-l 
C -H 

a id 
J5 o 

T3 Er> E 
•H 
O 

CO 

a; 

O CO 

•h ca 

o x, 

tj a. 
c x 

a; a> P. 

,a < — i 



Ch 

o 



O *H 






C 






•H O 






B £ 












1 O 








*H 
















c o 






'tH -P 












cu 






O X3 


O 




-P 


M 




o cc 


J3 










^•-t CD 








r. . 
-•— . 




i>s 'H 






ij C_, 


H 




•H O 


o 




O f~ t 


ft 1 










fx o 












CJ £ 


G 
















o c 






o 


CO 




C £ 


-T! 




O E 






•H 






4J ,— j 


— 




cc >> 






M -P 


n 




-p 3 






c ^ 


CD 




o cc 


X! 




a u 


4-> 




c -P 






CD 


O 




O -P 


M 




a? s 






-P 


CC 




ca «- 


c 




~ O 


o 


• 


P, * 


•H 




K O 






O 




o 




~ 


+^ 


C -P 


c 




•H 


o 


U 


C *H 5 


u 


c 


O 






CO 


T3 


CO 


-P Q) 


C 




O -P 




■ — 










CO 




^ o 


H 




Qi & 


C 


G 


CO 




co 


0) -P 


E 


£3 




>. 






CO 


ft 



I 



•H 



117 



CD<© 





O □ <D 




□ «© 



o 



o 

ID 



a> 



a> 

Q. 

E 



O 



O 
ro 



O 



Q 
ro 



Q 



ra a 

-m u 

■H 3 

-a -h 

C <m 

T3 O 

•H Cm 





a 



03 

o o 

■H £ 

O -P 

c w 

a) ca 

J3 







i a> 

& J5 
■P 

O 0! 

'•M M 

0! 

ca s 

O 03 
rH 
O O 
CO ^2 
Cm e 

>3 03 
-P 

•h o 
o .C 

CO -P 

a T3 

CJ c 
Q 

C 

O 03 



a; o 

M >H 

3 

+3 T-l 

E! 

m g 

o o 

p, o 

a m 

+5 0) 

E O 



O r-H 

O «s 
Cm 

O . 

m 

-M CD 

O -M 

a; -h 

Cm rH 
Cm 

w 



o 
CO 



I 



■H 



6 

■H 
O 
3 



CL 

CD 

X 

a> 



o 

N 

a 

CD 
& 
o 
(3 

s 

CO 

i a> 

a. co 
o 





■H 

h 

O 



o 

09 


+J 
O 

en 


>> ^3 
-l-> 



a) 



P< CD 
03 * 

CJ 



CD 
C 

o 



CJ 



o o 

•H O 
+» 



o 



C 

o 

O T3 
CD 

T3 -t- 5 
C 03 

3 

•1-3 
CO XJ 



0) 
CO 
Hi 
43 

— 



o 
s 

GO 

o 

£! 

EH 



o 

£ 

o 



3 

n 

od 
s 

B 
3 

■P 

Ih 
K 
Pi 
E 
CI) 



-P 
CJ 
B 
Eh 
<h 
O 



o 
>o 
CO 
-»-> 
CD 
B 



<u 

E- 



O 

CM 



' > I 



119 




120 



interest is shown in Figure 3-21 and demonstrates the separation 
achieved . 

Validation of the Ion-Pair Method 

Linearity and detection limits . Known amounts of PABA (0.13 - 1.0 
mg/L), PAHA (0.18 - 1.42 mg/L, PAABA (1.63 -1 13-04 mg/L) and PAAHA 
(3.23 - 25.86 mg/L) were injected in aqueous solution and subjected to 
the HPLC proceudre. All analytes demonstrated a linear relationship 
between peak height and concentration over the range of interest 
(Figures 3-22 and 3-23). The slope and corresponding standard error for 
the mean of 6 standard curves was 31.7 + 0.44 for PABA; 17.1 + 0.20 for 
PAHA; 17.7 ± 0.19 for PAABA and 8.5 + 0.06 for PAAHA. The consistently 
low standard errors were a result of the sample preparation procedure in 
which no extraction step was employed and errors due to extraction were 
avoided. This also serves as confirmation that no internal standard was 
necessary. Several compounds were investigated for use as internal 
standards, however, and anthranilic acid eluted between PAABA and PAAHA 
which made it a good candidate for this purpose. The conditions . of 
analysis had to be altered to completely resolve anthranilic acid, and 
this was accomplished at the expense of analysis time so it was decided 
to conduct sample measurements without the use of an internal standard. 

The limits of detection, taken as the peak height corresponding to 
twice the baseline noise, was found to be 0.06 mg/L for PABA; 0.12 mg/L 
for PAHA; 0.11 mg/L for PAABA and 0.24 mg/L for PAAHA. 

Recovery . Each compound, at the same concentrations as the aqueous 
standards, was added to pooled blank urine collected from fasting 
subjects under the test conditions outlined in the methods section. 
Recovery was determined by replicate analysis (n = 6) at each 



121 



100 



80 



60 



£ 

e 



*<D 

X 



a 40 

CL 



20 



0- \J 



-i i_ 



j i i i_ 



I L 



6 12 18 
Minutes 



ez: 




J I L 



6 12 18 



Fig. 3-21. Typical chronatograms under the conditions in table 2-4, of 
a pooled blank urine (left) and a spiked urine containing 
0.5 mg/L p-aminobenzoic acid (i), 0.71 mg/L p-aminohippuric 
acid (II), 6.52 mg/L p-acetamidobenzoic acid (ill), and 
12.93 mg/L p-acetamidohippuric acid (IV). 



122 




. 3-22. Typical standard curves of peak height versus concentration 
for p-aminobenzoic acid (closed circles) and p-aninohippuric 
acid (open circles). 



123 




(LULU) JL|6!8H >lD8d 



02 
■H 



124 



concentration. The mean slopes of the peak height versus concentration 
curves for spiked urine were compared to slopes from curves obtained by 
direct injection of aqueous standards. The results are summarized in 
Table 3-9- 



Table 3-9. Recovery of PABA and its metabolites by the ion-pair HPLC 
method 



Compound 


Concentration 
range (mg/L) 


Mean 
slope 


Standard 
error 


Recovery % 
(n = 6) 


PABA 


0.13 - 1.0 


30.0 


0.23 


95 


PAHA 


0.18 - 1.42 


16.7 


0.38 


98 


PAABA 


1.63 - 13.04 


17.7 


0. 13 


100 


PAAHA 


3.23 - 25.86 


8.4 


0.03 


100 



Precision . The between-day precision was assessed by analyzing the 
four concentrations of spiked urine pool for 6 days. The coefficient of 
variation for each compound was less than 1G$ for all urine concen- 
trations tested (Table 3-10) except for very low concentrations of PABA 
and PAHA. The within-day run variation in peak height was less than 5% 
CV at the lowest and highest standard concentrations for all compounds 
tested (n = 6). 

Selectivity . Good selectivity of the method under test conditions 
was indicated by the fact that no interfering peaks were observed in 
pooled blank urine (Figure 3-21 ). The y intercepts of curves 
constructed from spiked urine were 0.039 ±0.13 for PABA; -0.66 + 0.32 
for PAHA; 1.78 ± 0.98 for PAAHA and -0.059 + 0.50 for PAAHA (n = b) 



125 



Table 3-10. Between-day precision of the ion- pair HPLC method 





Mean concentration 


C.V. % 


^ u ui p u u n a 




In - 6; 


PABA 


0.98 


3.0 




0.48 


9.2 




0.25 


9.2 




Kim I J 


10./ 


PAHA 


1 .40 


3.6 




0.67 


2.3 




0.32 


5.4 




0.18 


14.0 


PAABA 


12.4 


6.5 




6.5 


3.8 




3.3 


3.2 




1 .6 


8.3 


PAAHA 


26.0 


2.9 




12.8 


1.4 




6.5 


1.4 




3.3 


5.3 



also indicating no significant interferences. Blank urine from non- 
fasting subjects who were not restricted from drug use, however, did 
demonstrate many interfering peaks which suggests that test conditions 
must be rigorously controlled for good selectivity. 

Evaluation of Bentiromide Metabolism 
in Clinical Samples 

The development of the ion-pair HPLC method now permits the study 
of individual metabolite concentration patterns in clinical samples for 
the purpose of detecting false positive bentiromide test results. In 
this section, the appropriate urine collection interval will be esta- 
blished through pharmacokinetic studies. The metabolite concentration 
patterns in normal subjects, pancreatic patients and liver patients will 



126 



be determined under the established conditions and evaluated for signi- 
ficant differences. Additionally, various parameters extracted from 
these metabolite concentration patterns will be evaluated for their 
effectiveness in the differentiation of false positives due to liver 
dysfunction. 
Pharmacokinetic Studies 

It is reasonable to assume that PABA metabolite concentration 
patterns will vary during the 6 hour collection interval normally used 
for the bentiromide test. It is also possible for certain metabolites 
to reach their peak concentrations in urine at different times after a 
dose of bentiromide has been administered. Thus it is important to 
investigate this possibility so that potential differences may be 
optimally determined. The concentration versus time profiles for each 
metabolite quantified from normal volunteers is presented in Figure 
3-24. This experiment was conducted according to the protocol outlined 
in the methods section of this work and quantitation was achieved by the 
HPLC method. The concentrations of PABA and PAHA are too low to be 
useful in the normal subjects tested although it is possible that 
analytically useful concentrations may be present in hepatic patients. 
The concentrations of PAABA and PAAHA were high enough to be useful and 
demonstrated concentration maxima at 2 and 3 hours, respectively. If 
PAABA and PAAHA concentrations are to be used for the evaluation of PABA 
metabolism in a single sample, a urine collection interval must be 
chosen that would best reflect any metabolic lag time. The collection 
interval of 0-3 hours was initially chosen for single sample analysis 
because metabolic lag might best be detected prior to the concentration 
maximum. Also concentrations would be high enough for reliable 



127 



80 



60 



e 



"cu 

L. 

o 

X 

UJ 



40 



o 

I 20 



-10 




2 3 
TIME (Hours) 



i 

5 



■ 

6 



Fig. 3-24. 



Urinary excretion versus time curves for p-aminobenzoic acid 
and metabolites (n=5). See figure 5-10 for a definition of 
symbols and table 2-4 for chromatographic conditions. 



128 



quantitation over this interval. The large standard deviation 
represented in Figure 3-24 is reflective of individual variation in the 
subjects tested and indicates that metabolite concentrations in liver 
patients must be widely variant for differentiation. 

An attempt was also made in this study to evaluate the 
pharmacokinetic mechanism of PABA metabolism. Although no meaningful 
data could be obtained for PABA and PAHA, the data in Figure 3-24 
suggests that PAABA is eliminated in approximately first order fashion 
whereas PAAHA shows zero order elimination. This suggests a metabolic 
conversion of PAABA to PAAHA and direct renal excretion of PAAHA without 
further conversion. To prove that this is the predominant mechanism of 
PAAHA synthesis, however, the rate of formation of PAAHA must be 
approximately the same as the rate of elimination of PAABA. This study 
is not possible in urine samples, however, since PAABA is significantly 
excreted into urine and its elimination must be due, at least partially, 
to this mechanism. 

Another factor which supports the idea that PAABA is further 
metabolized is that a peak resulting from PAABA glucuronide formation 
appeared in the chromatograms of samples collected more than 3 hours 
after administration of bentiromide. The peak was confirmed as PAABA- 
glucuronide by treatment with glucuronidase which eliminated the extra 
peak and caused an increase in the PAABA peak height (Figure 3-25). 
This observation does not support the claim that PAABA is converted to 
PAAHA, however, and all that can be said about metabolic mechanisms is 
that PAABA appears to be further conjugated whereas PAAHA appears not to 
be. 



Fig. 3-25. Typical chromatograms under the conditions in table 2-4 of a 
urine sample collected at the 3-6 hour interval from a 
subject undergoing the ben tironide test. The chromatogram 
on the right is from the urine sample after treatment with 
glucuronidase and the one on the right was untreated. The 
presence of p-acetamidobenzoyl glucuronide is indicated by 
the disappearance of the shoulder on the p-acetamidohippuric 
acid peak (il) and the increase in peak height of the p- 
acetamidohippuric acid peak (i) after treatment with 
glucuronidase. 



131 



Analysis of Clinical Samples 

The PAABA-glucuronide peak was not resolved from the PAAHA peak so 
it was decided to modify the sample preparation procedure of the HPLC 
method to include a glucuronidase treatment step. The optimum time 
interval for incubation of samples with glucuronidase was determined by 
monitoring metabolite peak heights versus time of incubation. The 
results of this study are presented in Figure 3-26 and show an increase 
in PAABA peak height accompanied by a decrease in peak height of the 
unresolved PAAHA-PAABA-glucuronide peaks. Figure 3-26 also shows an 
increase in PAHA peak height that suggests formation of PAHA 
glucuronide. The optimum incubation time was chosen as 30 min because 
that was the minimum time for stabilization of all analyte peak 
heights. The specificity of the glucuronidase enzyme with respect to 
analyte species was also validated by this study since no changes in 
peak heights were observed after 30 minutes of incubation with an excess 
of enzyme. 

All clinical samples were analyzed by the modified procedure so 
that an accurate assessment of both acetylation and glycine conjugation 
of PABA could be made. The amount of each analyte excreted over the 
initial three hour collection period in normal subjects, pancreatic 
patients and liver patients is presented in Table 3-11. The extremely 
low amounts of PABA and PAHA excreted and the broad range of values 
within each population made these results useless for the 
differentiation of metabolite synthesis. This was proven by statistical 
evaluation in which no differences were found by comparing population 
means even at the 50,^ confidence level. The results from PABA analysis 
were evaluated by a non- parametric procedure ( 1 63) since the 



409 



30 



e 

£ 



an 



-□ — □ — □ — □- 



-□ 



I 20 



o 

CL 




O O O o 



20 30 40 
II ME (Minutes) 



26. Plot of peak height versus time of treatment with 

glucuronidase for p-aminobenzoic acid and its metabolites 
The chromatographic conditions are listed in table 2-4 an 
the symbols are the same as for figure 3-10. 









CO 


I- 






^> 


P 


• 


• 


• 






to 




o 


r — 




E 
























< 










3 












O 






o 




o 




-*i 






a 




K> 


a. 


id 










£3 


r<"\ 






1 

O 












m 












h 













CD 




p 








> 




to 




'O 




c 


E 






















m 












0) 


-a! 










rH 


pq 












< 










E 


< 


C 


cc 


O 




CO 


a. 


CO 






m 


co 




0) 




CO 








B 




CNJ 


o 


rH 












CO 












o 












•H 












c 








c 




■H 






rO 






rH 




aj 




• 


• 


O 






LT\ 


1 








c 


1 




a 


E 


ED 


C\J 


gp 


6 


■H 








CNJ 














• 


~ 


•=: 




d 




o 


0) 


X 










-t-> 


<! 










o 


a. 










(h 












u 




c 








X 










O 






1) 




• 


• 















CO 



o 

cd 
-p 

CD 
E 

co 
-p 



C XJ 
!0 O 
•H 
<S J* 

« a) 
< a. 



a. 



C 
c 



-p 

-P O 

3 rH 
O rH 
E O 

< a 



6 



CM 

c 



c ° ° 



CO 

6 



a 






CO 


z 


C\J 




LTV 


0) 




• 




g 


d 


O 


d 



I 



a 

Eh 









o 











■H 






■H 




-P 




-p 


-P 




a) 




a 


a 


rH 


0) 




<u 


rH 


C 


Sh 






2 




o 


o 


^3 


P. 


z 


c 


> 


3 


O 


a 


d 


■H 


CO 


(X 


a 


P, 


rH 



134 



distribution could not be assumed normal because of skewness by results 
of zero. The results for PAHA, however, were normally distributed and 
evaluated by the parametric t statistic (149). 

The results from PAABA and PAAHA analysis showed promise for 
metabolite differentiation and suggest that there may be a larger amount 
of PAABA excreted in liver patients than in either normal subjects or 
pancreatic patients. However, the large standard deviations observed 
rendered these differences insignificant even at the 50$ confidence 
level and an alternate strategy was necessary to demonstrate possible 
differences. Low total recovery of metabolites is expected in 
pancreatic patients and some liver patients and this would tend to mask 
individual metabolite differences if only one metabolite was 
evaluated. A ratio technique was employed, therefore, to correct for 
variations in total recovery. Since relatively high values for PAABA 
excretion in liver patients was suspected, the ratio technique used, 
expressed the ratio of the amount of PAABA excreted to the total amount 
of PAABA + PAAHA. The mean ratios for normal, pancreatic and liver 
patients were 0.622+0.114, 0.692+0.113 and 0.738+0. C69, respectively. 
Again, a comparison of these results by the students t test showed 
insignificant differences although it did appear that there may be an 
increase in PAABA relative to the amount in liver patients. 

At this point, it was decided to evaluate the situation from a 
pharmacokinetic point of view to see if the sample collection protocol 
could be altered to demonstrate a significant difference in PABA 
metabolism among the three groups. It was observed that significant 
variations occurred in the time at which PAABA and PAAHA reached their 
maximum levels in individual subjects during the pharmacokinetic study 



135 



presented in Figure 3-24. This adds to the variation found in PA ABA and 
PAAHA excretion for the 0-3 hour collection interval since sample 
collection was halted near the time of maximum excretion. Also, erratic 
results were observed for the 0-3 hour interval in a study to determine 
the formation rate of PAABA and PAAHA (Figure 3-27). In this 
experiment, a longer collection interval and shorter sampling times were 
used to provide more experimental points for pharmacokinetic analysis 
but the erratic excretion levels found between 0-3 hours precluded such 
an evaluation. The results do suggest, however, that there is a greater 
variation in PAABA and PAAHA excretion during the first three hours 
after bentiromide administration. Apparently, a steady-state of 
physiological hydration of the subject which provided a more consistent 
elimination rate of PABA metabolites is reached after the 0-3 hour 
interval. Consequently, it was thought that a collection interval of 3- 
6 or 0-6 hours might be more appropriate for metabolite evaluation and 
studies were carried out to determine this. Samples collected between 3 
and 6 hours after bentiromide administration were also collected during 
the clinical study and were quantitated to evaluate their usefulness in 
metabolite differentiation. The results of this study are presented in 
Table 3-12 and show that there is slightly less variation in metabolite 
excretion within each study group as indicated by lower relative 
standard deviations. Statistical evaluation of individual metabolite 
excretion again showed no significant differences at the 50^ confidence 
level among subject populations. Amounts of PABA and PAHA excreted were 
evaluated by non- parametric methods in this case and PAABA and PAAHA 
were evaluated with the parametric t statistic. The ratio of PAABA to 
total PAAHA + PAABA in these samples was 0.247+0.046 for normal 



136 



40 



30 



E 



O 



L. 

(J 
X 

UJ 



o 
c 



20 





10 



1 2 3 4 5 
TIME (Hours) 



Fig. 3-27. Urinary excretion versus time curve for p-aminobenzoic acid 
and its metabolites from one subject over an extended time 
period. See figure 3-10 for a definition of symbols and 
table 2-4 for chromatographic conditions. 



u 

o 

X. 
I 







to 








p 




• 


• 


Etc 


CO 








G 




















< 






















a 




o 




< 


d 


• 


• 


• 




o 




o 


00 




S 




LP 





u 
o 
> 
o 

01 

0) 



u 

■H 
C 



c 

■H 

a 
-t-> 

0) 

a 
x 
o 

CO 





lf\ 


CO 




p 


• 


• 


• 


w 






c 


B 


CO 


cc 


VO 






• 




Hi 




o 


CM 


£ 




























• 


a 






K\ 


EC 


• 




1 






CM 


CM 




6 


1 

o 


Z 


c 


cm 


CO 


to 


EG 




; — 




a 




• 


• 


E 


d 


o 





c 

It 

o 
s 

CO 
4- 
■H 

i — I 
TJ CO 

c > 

CD 

< 4J 

P3 s 
-a; T-t 

c 

Vi o 

O -H 

+J 

-P o 

a <z> 

3 i-H 
O i-H 

E O 
-a; O 









1! 






be 






a 


C\J 




a 








i 




C 

EO 


o 
r- 









• 


s 


o 


O 



O 



CM 



I 



a 





c 




CJ 






c 




•H 






■H 




+J 










cd 




o 


.-3 


i-H 


3) 




CJ 


i-H 


a 


Ph 




■( — ; 




E 


CJ 




~ 


P- 


S-l 


c 


> 






O 


CO 


•H 


CO 


P. 


c 


p. 


i-H 



138 



subjects, 0.556+0.125 for liver patients and 0.402+0.248 for pancreatic 
patients. Although there was a significant increase in the ratio 
between normal subjects and liver patients (p<0.00l), the difference 
between both normal subjects and pancreatic patients and between liver 
and pancreatic patients was not significant. This is because of the 
large variation observed for pancreatic patients and is probably due to 
slow PABA recovery resulting from deficient enzymatic cleavage of 
bentiromide. This would cause an increase in the time of maximum 
metabolite excretion in these patients and thus cause a shift in the 
variablity associated with this factor. 

A 0-6 hour collection interval was simulated by adding the amounts 
of each metabolite excreted during the 0-3 collection interval to those 
from the 3-6 hour interval and the results are presented in Table 
3-13. Again, there were no significant differences (p>0.5) between 
individual mean amounts of metabolites excreted, among the study 
groups. The amounts of PABA and PAABA excreted were still too low for 
evaluation and highly variant so these two compounds were deleted from 
further consideration. The ratio technique when applied to the 6 hour 
data, however, demonstrated significant increases of 0.220 (p<0.00l) and 
0.211 (p<0.00l) for the liver patient population as compared to the 
normal and pancreatic populations, respectively. The mean values of the 
PAABA to total PAABA + PAAHA were 0.427+0.118 for normal subjects, 
0.647+0.120 for liver patients and 0.436+0.100 for pancreatic 
patients. There was no significant difference between the populations 
of normal and pancreatic subjects (p>0.5), however, which indicates that 
the ratio technique was successful in correcting for variable total 
recovery. 



139 



0) 
rH 
ft 



CO 
03 



3 

O 



I 

o 



M 

1) 
> 

o 

to 
o 

H 

P. 
E 

.1! 
CQ 



9) 
O 



W 
O 
-P 
•H 
rH 
O 
& 

to 
■(J 
o 
s 

CQ 
4J 



TD CO 
C > 
CO ^ 
0) 

■< 

CP c 

=C -H 

ft 

c 

tM O 
O -H 
-P 

+J O 
C CU 

3 rH 
O rH 

E O 
■a; O 















• 




w 


CT> 








CM 




CM 




— 






CO 








cu 




'O 




E 












CM 




O 




• 




w 




CM 








to 


LO 


c 




to 




CO 


• 


• 


* 


CU 






to 


E 


5 


m 


00 



c 












•H 






to 






H 




0) 






rQ 







u 


L.O 


oo" 

1 








c 




«* 


a 


E 


cd 


CM 


oo 




•H 




h 


• 


CM 


r- 

• 


T3 


^ 




O 




o 


d) 












-P 












0) 


ft 










~ 












OX 




an 


to 

to 




>* 






u 




• 


• 






2 


C\J 


to 


CM 



I o o 



LO 





CO 




to 




cc 


CM 






• 






>- 




CM 




so 
u 


6 


1 

5 


• 

o 


c 






CM 


a 








cu 








2 


d 


d 





to 

I 

to 
o 

rP 

cc 
Eh 





3 















• H 






•H 




4-> 




+J 


+a 




CO 




u 


cd 


rH 


0) 




cu 


i — 1 










3 




c 




,o 


ft 




c 


> 


3 


c 


O 


CO 


•H 


CO 


a, 


c 


ft 


— i 



140 



These studies indicate that liver patients can be distinguished 

from both normal subjects and pancreatic patients in clinical 

situations. They also show that only PAABA and PAAHA need be 

quantitated for this purpose and that a 0-6 hour collection interval 

with the ratio technique must be used. The diagnostic implications of 

this study will be discussed in Chapter 4 of this dissertation. 

Analysis of Bentiromide Metabolites 
by Room Temperature Phosphorescence 

Throughout the studies on PABA and its metabolites by HPLC, PABA 
and PAHA were determined to be of little use for the differentiation of 
patient populations. The major metabolites PAABA and PAAHA, however, 
are useful for this purpose and will be the focus of the work described 
in this section. Quantitation of PAABA and PAAHA in clinical samples 
requires that measurements be made in a mixture of all metabolites 
unless a separation step is included. Therefore, the room temperature 
phosphorescence characteristics of all metabolites must be studied to 
adequately assess the analytical possibilities and provide the most 
efficient analysis system. The goal in this section will be to 
differentiate the room temperature phosphorescence of these metabolites 
from one another and various methods will be investigated for this 
purpose. If metabolite differentiation can be incorporated into the RTP 
analytical format, a convenient and rapid method for bentiromide test 
evaluation which is not subject to false positive tests in liver 
patients may be provided. 
Spectral Studies 

The most obvious way to attempt to differentiate the 
phosphorescence of several compounds in a mixture is by spectral 



141 



discrimination. If the RTP emission maxima of PA ABA and PAAHA were 
sufficiently separated from PABA and PAHA and from one another, emitted 
radiation could be monochromated to filter out emission from all 
compounds except the one of interest. This possibility was investigated 
by determining the excitation and emission spectra of each compound 
individually. The absorption spectrum of each was also determined and 
the absorptivities in aqueous solution were calculated. The results of 
these studies are presented in Table 3-14 and the RTP emission spectra 
on the DE-81 substrate are shown in Figure 3-7. Table 3-14 and Figure 
3-7 show that the feasibility of spectral resolution of these four 
compounds is slight because of the small separations between their 
emission maxima and the breadth of the emission peaks. The fact that 
the RTP intensity of PABA was more than 10- fold that of either PAKA or 
PA ABA, and more than 100-fold that of PAAHA, may prove to be useful for 
individual quantitation of metabolites if a separation is employed. For 
example, if there were some way to remove PAABA from the sample and not 
PAAHA, the relative quantities of these metabolites could be determined 
by difference. This could be accomplished by analyzing the sample by 
the RTP procedure for total metabolites and then analyzing the separated 
sample in the same manner. The difference between the two measurements 
would reflect the relative amounts of PAABA and PAAHA. Since the 
amounts of PABA and PAHA excreted are negligible, the ratio of the 
signals from separated and nonseparated samples would be analogous to 
the ratio shown in the previous section to be significantly different in 
liver patients than in pancreatic patients, and could potentially be 
used for diagnostic purposes. To preserve the convenience of the RTP 
method, however, it would be advantageous to obviate a separation step, 



142 



to 
a 



* 








I — 1 


o 






PL, 


r>"N 


O 




Ph 









■si 
a 
o 

■H 

■P 

3 

rH 
O 
K 



O 

a 
a 

5 
03 
•H 



Ph 


£ 


CO 


Eh 


C 


CM 




-< 


"ST 





E 




o 


a 









a 
o 



o 

X 



p. 

o 
m 



-a 
c 

O 

E 
c 
u 



LTv 
CM 



a, 




«*■ 


o 




cu 


CJ^ 


c 




-< 


CM 





CM 



CM 











CNJ 






CO 


• 






• 



o 


O 


CM 


in 








o 


CM 


cm 


CM 


CM 







< 


< 


< 


< 


pq 


S3 


pq 






< 


< 


< 


<i 


< 


Ph 


Ph 


a, 


Ph 



<-> 
















CO 




* 












s 




o 




fn 




"in 




a 




u 




o 




03 




c 




•H 




■P 




•H 




03 




c 




CU 




+» 




c 




■H 




CU 




o 




c 




a; 




o 




ea 




0) 




Sh 




o 




J3 








03 


• 


o 






c 


p< 






O 


(1) 


PV 


> 


E 


•rH 


o 




a 


a 




rH 


X! 


CD 


O 


Fh 


a 




a 













o 



143 



and other means were investigated to accomplish the analysis without 
separation. 

Selective Heavy Atom Effects 

The convenience of the RTP method would not be significantly 
affected if the selective enhancement of phosphorescence by a heavy atom 
perturber could be used. Ideally the phosphorescence of either PAAEA or 
PAAHA would be selectively enhanced by a particular heavy atom species 
making it possible to correlate the ratio of PAABA to PAAHA concen- 
tration with the ratios of phosphorescence intensity of the two with and 
without the heavy atom present. Three heavy atom species were studied 
for potential selective enhancement effects on PABA and its metabolites 
and the results are shown in Table 3-15. The iodide ion enhanced the 
phosphorescence of all test compounds and the degree of enhancement was 
3.16-fold for PABA, 4.92-fold for PAHA, 2.90-fold for PAABA and 6.34- 
fold for PAABA. Tl + and Ag + enhanced the phosphorescences of the 
individual analytes to about the same degree with the greater 
enhancement observed for PAABA. The approximate degree of enhancement 
for both heavy atom species was 1.75- fold for PAHA, 3.52-fold for PAABA 
and 3. 38- fold for PAAHA. The phosphorescence of PABA, however, was 
apparently decreased 1.77-fold in the presence of these two heavy atom 
species. 

For the purpose of distinguishing PAABA from PAAHA by selective 
enhancement in the complex mixture represented by the sample, several 
factors must be considered. The choice of heavy atom depends on the 
degree of selective enhancement of phosphorescence required for either 
PAABA or PAAHA, the enhancement of phosphorescence of PABA and PAHA 
which would potentially interfere with the measurement and the 



144 



CD 
O 

O 
03 
o 
H 
o 

-C 

O, 
ra 




a 

M 
3 
4J 

m 

P- 



E 
O 

c 

Eh 

CD 

-P • . 

LP* 
C H 

o a 

CO 

0) CO 

•H CD 

O -P 

<L> -H 

Ph rM 

CO O 

E CO 

O -P 

-P 01 

CO E 



CO 



>. 

> ■ 

CO -H 
0) 

C 

r-H a 
CO 

u < 

O PQ 

> «C 
O Ph 
CO 

Cm 
Cm O 



o 
O 
Cm 

cu 



01 
CD 
■H 

■H 
•H 

tn 
c 

0) 

-p 
c 

■H 



iO 
O 



"3 



PS 



c 

co 

E 



O 



E-i 







h 






CO 




Ph 


1! 


• 


OS 


E 


o 










Q 






K 








gr 










C 


• 


1—1 


a 


c 


Oh 


CD 






E 



o 

C\J 



0-. 1) 
« E 



E 

1 



L— 


ro 








• 


• 




*~ 






•"- 












LO 




LP 


C\j 




c- 


CM 












LP. 










• 


• 


















CM 








C\J 


c 


oo 


C\J 










(J » 




-C 






• 




• 


CO 




CM 


CM 


to 






o 


c 




LO 




o 


Lf> 






f*. 












t<"\ 


O 




33 






• 


■ 


• 











8 



I 

f>~\ 



E 



O 

LPv 



E 



LP. 



CM 
LTi 



X 


<* 


c 




CD 




o 


r- 


r< 


C\J 




C\J 



c- 



°2 


LPi 


— 




C\J 


fPi 






-3" 



C\J 



E 

O 



CO 
E-< 



Ph PU 



<5 

Ch 



145 



enhancement of phosphorescence of endogeneous components of urine. It 
appears that the best approach would be to use iodide since the effect 
on the phosphorescence of PAAHA was more than twice the effect on that 
of PAABA . However, calculations based on the average concentrations of 
PAABA and PAAHA found in patients suffering from liver and pancreatic 
diseases by HPLC, demonstrate that the percent increase in total 
phosphorescence intensity from all analytes due to PAAHA enhancement 
would be only 15.1$ for the pancreatic disease population and 1.0% for 
the liver population. Although this reflects the higher relative amount 
of PAAHA excreted in the pancreatic population, the difference of 8.1$ 
is within the limits of experimental error established for the RTP 
method. This small difference is due to the relatively low contribution 
of PAAHA phosphorescence to the total phosphorescence and negates the 
analytical utility of this approach. 

The use of Tl + or Ag + to selectively enhance PAABA with respect to 
PAAHA would be a better choice if the fraction of total phosphorescence 
was the only important factor because PAHA and PAAHA phosphorescences 
are not enhanced as much as by iodide and the phosphorescence of PA3A is 
decreased. However, the percent increase in total phosphorescence due 
to selective PAABA phosphorescence enhancement would be only 0.4% higher 
in the liver patient population than the pancreatic patient population 
because of the small difference in enhancement between PAABA and 
PAAHA. Although it was not studied because of the small differences 
found in these initial experiments, the enhancement of endogeneous 
urinary components would lower these differences by increasing total 
phosphorescence and it appears that selective enhancement is not a 
viable means to differentiate PAABA and PAAHA in clinical samples. The 



146 



observations made here are valuable in characterizing the physical 
aspects of the phosphorescence of these compounds and a discussion of 
these aspects will be included in Chapter 4. 
Lifetime Studies 

Time resolution is another potential means of distinguishing 
between PAAEA and PAAHA without a physical separation step. If the 
phosphorescence lifetime of one of these components is sufficiently 
longer than those of the other components, phosphorescence measurement 
can be made after a specified delay time in which the phosphorescences 
of all other components have vanished and the signal is proportional to 
only the concentration of the longer lived component. This is an ideal 
situation, however, and such lifetime differences are not commonly 
observed for structurally related compounds such as PABA and its 
metabolites. It may be possible, however, to relate relative 
concentrations of the compounds of interest through a study of 
phosphorescence relative intensities and lifetimes with respect to their 
expected concentrations. The phosphorescence lifetimes of PABA and its 
metabolites spotted from aqueous solution onto untreated and iodide 
treated DE-81 filter paper are presented in Table 3-16. These lifetimes 
were determined from phosphorescence decay curves in which 
phosphorescence intensity was measured versus time after cutting off the 
exciting light. The lifetime was calculated by regression analysis of 
the natural logarithm of phosphorescence intensity versus time curve and 
the lifetime was taken as the inverse of the slope of this curve. The 
lifetime determined in this way is the time it takes for the 
phosphorescence intensity to decrease to a fraction (1/e) of the initial 
phosphorescence intensity. 



147 



0) 




u 




f-l 
M 












03 




w 




□ 














T3 




CD 


G 


T3 










CQ 




C^ 




"\ 












O 






< \ 


at 


* 










p: 
. — i 




03 
















>- 




CO 




■ 












Ph 




Eh 




o 




M E 




0) o 




E -P 




•H CO 






G 


0> >> 


o 


Vh > 


*j 


•H CO 




r— 1 0) 




rC 


>s 


1) 


> 


O V) 


EC 


c o 




o 




O CD 




03 O 


O 


CD C 




w CD 




O 03 








a. a 




03 




o "a 




~ C 




Ph CD 





I 



TO 
E- 



H 



CO 








z- 


cr 




o 


o. 


cr 






• 




• 


• 


o 


d 


o 


O 



C 



C- 







03 


c 


E 


3 




<D 







ITv 

c 



o-. 



3> 



c 



8 - 

• • 

o o 



CD 

s 



CM 



CM 

CT. 
CTv 



o 





CM 


c 


to 




p 


o 


o 


o 


o 


CO 


• 




• 








c 


o 


G 





C\J 




oo 










co 










C- 






<* 


cr 







«< 




< 




S3 








< 


< 


< 


< 


<< 


< 


Ph 


Ph 




Ph 



CD 
> 

u 
a 

u 

CD 
E 



03 
3 

> 
H 

ps 



CJ 



Cm 
O 



c 

01 



Cm 
03 

o 
o 

a 

o 

■H 
+J 

as 

iH 

!h 

M 

o 

CJ 

CJ 



C 
CD 

i— I 
> 
CO 

+> 

03 
■H 

M 



148 



The two possibilities for time discrimination of PAABA and PAAHA are 
to gate out the phosphorescence of PAABA in the absence of a heavy atom or 
to gate out the phosphorescence of PAAHA in the presence of a heavy atom, 
since the lifetime of PAABA is relatively shorter for the former and 
longer for the latter case. The lifetimes of PAABA and PAAHA are 
different by a factor of only about 2 in each case, however, and com- 
pletely gating out the phosphorescence of one analyte would result in low 
signals from the second. The interference from PABA and PAHA must also be 
considered in each case since their lifetimes are also similar. 

Consider first the temporal resolution of PAAHA from PAABA in the 
absence of a heavy atom at a gate time of 1 .8 sec. The phosphorescence 
intensities of PABA, PAHA, PAABA and PAAHA would be reduced by a factor 
of (1/e) 4 ' 6 , (1/e) 3 ' 2 , (1/e) 4 ' 1 and (1/e) 2 ' , respectively, at that gate 
time. Again, if the resulting RTP intensities are based on a calcula- 
tion from RTP standard curves involving the RTP intensities from mean 
concentrations of analytes found in liver and pancreatic patients, the 
total phosphorescence intensity would be reduced 94.9$ in the pancreatic 
patient population and 36.0% in the liver patient population. The 
higher signal calculated for the pancreatic patients reflects the higher 
relative concentration of PAAHA although the difference in populations 
is only 1.1$ and well within the range of error for the method. 
Shortening of the gate time would only decrease the difference between 
populations and a longer gate time would result in total phosphorescence 
intensities that were too low for reliable analytical measurements. 

Similar calculations were carried out for the temporal resolution 
of PAABA from PAAHA in the presence of a heavy atom and at a gate time 
of 0.2 sec. In this case, the mean total phosphorescence intensity of 



149 



PABA and its metabolites would be decreased 71.8% for pancreatic 
patients and 77.1$ for liver patients which again reflects higher 
relative PAABA excretion in the liver patient population. The 
difference of 0.7$ is not useful analytically and an alternate strategy 
was necessary to incorporate metabolite differentiation into the RTP 
format. 

Differential Hydrolysis of Metabolites 

It has been shown in studies represented by Figures 3.4 and 5.5 
that acid hydrolysis of PAABA to PABA is 95% complete after 15 minutes 
whereas acid hydrolysis of PAAHA to PABA is only 4% .complete. Alkaline 
hydrolysis, on the other hand, is complete for both compounds after 1 
hour. This difference in hydrolysis characteristics of PAABA and PAAHA 
can be used to differentiate these metabolites by carrying out separate 
acid and base hydrolysis procedures on aliquots of the same sample. The 
differences between phosphorescence intensities of these hydrolysis 
mixtures should correlate with the differences in relative PAABA and 
PAAHA concentrations found in subject populations. This correlation is 
made possible by the fact that liberated PABA demonstrates much higher 
phosphorescence intensity than its metabolites and the total 
contribution to the phosphorescence signal from unconverted metabolites 
in the acid hydrolysis mixture would be small. Thus the ratio of total 
phosphorescence intensity measured from PABA in the acid hydrolysis 
mixture to that of the alkaline hydrolysis mixture should approximate 
the ratio of PAABA concentration to the total PAABA plus PAAHA 
concentration measured by HPLC. 

The degree of interference from metabolites of PABA in the acid 
hydrolysis procedure after 15 minutes at 100°C can be estimated by 



150 



calculating the relative phosphorescence intensities that would result 
from the concentrations of metabolites that were determined by HPLC in 
that mixture. These concentrations were 10 mg/L of PABA, 286 mg/L of 
PAHA, mg/L of PAABA and 27 mg/L of PAAHA which resulted from acid 
hydrolysis for 15 minutes of 451 mg/L of PAAHA. The total 
phosphorescence contribution from these concentrations is only 17.6$ of 
the phosphorescence that would have been produced if all of the PAAHA 
had been converted. This would thus, provide a difference between acid 
and base hydrolysis of 80.4% (allowing for incomplete PABA recovery) in 
the amount of apparent PABA liberated from PAAHA. These calculations 
were based on the use of the DE-31 substrate without heavy atom 
treatment because all heavy atom species studied enhanced the 
phosphorescence of PAHA and PAAHA more than that of PABA. 

These calculations indicate that the method has the best potential 
for discrimination of PAABA and PAAHA of all the methods studied and a 
study of clinical samples was carried out to evaluate the method. The 
alkaline hydrolysis was carried out on clinical samples in the usual 
manner and the acid hydrolysis was stopped after 15 minutes by neutrali- 
zation of the mixture with 2 M NaOH containing 0.01 H KH 2 P0 4 . The 
results of the determinations in these hydrolysis mixtures of clinical 
samples is presented in Table 3-17. The relative phosphorescence 
intensities from acid and base hydrolyses appears to reflect the 
increased amount of PAABA excreted by liver patients since this ratio is 
higher than that obtained from both the normal and pancreatic patient 
populations. All of the ratios are higher, however, and the difference 
is not as great as for ratios calculated from HPLC data. This was 
expected since the phosphorescence from unconverted metabolites of PABA 





CP 




o 






o 


CO 


p 


CD 






u 


CO 










a 






• 




_ 


+j 


o 




o 


o 


o 


3 


O >H 










u 


-P 










CD 


CO 










ft 


T3 m 












■H 












a 


a 










CO 






■ 


• 








o 


o 


o 


E 












o 












o 












M 












>5 CO 












,Q -H 












to 


CO 










co >j 


•H 




o 




LPs 


CD 1 


' — i CO 


P5 








i-l o 


6C >-. 


12 


CO 


LTv 


Lf\ 


ft m 


E H 




C\J 






E T3 


O 










CO >> 












to ,q 


•< 13 












PP >> 










r— 1 CD 


•< -C 






CO 


CM 


ca c 


Ch 


c 


• 




• 


O -H 




Co 






CM 


•H rH 


■H 






LTS 




C CO 





s 






•H 


CO 










rH rH 












O CO 












C 13 












■H C 












CO 
























cd o 












M -H 


CQ 










3 T3 


■H 




o 






CO -H 


CO 


~- 








CO o 


>s 






LT> 


0> 


0) CO 


i-H 




LO 


fr>i 


LP 


E 


^ O 






















«: c 


E T3 












' >2 










< s 












Ph o 














P3 CD 










4-> 1— 1 


«; c 










C O 


Pu -H 








^ 


CD 


i — ( 










U 


d 




o 






a >-. 




CD 




CO 


CT 


ft u 


rH 


E 




& +> 


cC 










CO CD 












E 
























O M 












o 
























c a, 




C 


o 


o 


in 


3 CO 












o o 












E Si 












-X, ft 












• 




C 




o 








o 




•H 




1 




rt 




+> 






+J 




CO 




a 


CJ 


CO 


1— 1 


CD 




D 


H 


a 




h 


1 — 1 






E 


o 




f> 


^> 


P- 


fn 




> 


CO 


D 


O 


o 




■H 


Eh 


c/3 




C3 


ft 


rH 



152 



in the acid hydrolysis mixture would interfere with the quantitation 
PAABA by adding to the observed signal. Statistical analysis of the 
data shows that the differences among all populations for acid versus 
base hydrolysis are not significant even at the 50% confidence level. 
The differences between ratios are somewhat more significant, however 
and the liver population demonstrated ratios that were different from 
normal and pancreatic patient populations at 90% and QQ% confidence 
levels, respectively. The difference between normal and pancreatic 
populations was again insignificant at the 50% confidence level. 



CHAPTER 4 
DISCUSSION 

Potential sources of error that are associated with the bentiromide 
test as it is presently administered have been treated separately in 
this work. The problem of lack of analytical specificity was improved 
by the development of the RTP method for PABA analysis. Physiological 
non-specificity with regard to liver patients was also treated by the 
development of the liquid chromatographic method for PABA and its 
metabolites and the evaluation of metabolite excretion patterns in 
clinical subjects by this method. An RTP method was also developed for 
the differential analysis of PABA metabolites and evaluated with 
clinical samples. 

In ti. s section, a detailed discussion of the analytical merits of 
each method will be presented. Also, the variables involved in 
analytical development will be addressed from a mechanistic point of 
view and the diagnostic value of the methods will be evaluated in 
detail. During the course of this study, information relating to the 
physical aspects of room temperature phosphorescence of PABA and its 
metabolites was obtained and will be discussed here. 

Room Temperature Phosphorimetry of p-Aminobenzoic Acid 
The choice of substrate for PABA analysis was based on drying 
times, the dependence of RTP signals on pH, relative detection limits, 
and commercial availability. The DE-81 substrate was superior to the 
DTPA impregnated S&S 903 substrate in all categories except for 



153 



154 



drying characteristics. Currently accepted RTP theory suggests that 
rigid binding of the analyte to substrate material is one of the primary 
factors contributing to high phosphorescence intensity (94). So far, 
hydrogen bonding is believed to be the main mechanism responsible for 
the rigidity factor in filter paper substrates (44,94). This hydrogen 
bonding occurs between analyte molecules and hydroxyl groups present on 
cellulose in the paper substrate. The relatively high phosphorescence 
intensities observed on the DE-81 substrate are probably due to a 
combined effect of hydrogen bonding and ionic bonding between the ionic 
exchange groups on the DE-81 substrate and analyte molecules. This 
combined effect results in a more rigid binding of the analyte molecule 
to the substrate than for substrates which do not possess ion exchange 
groups and thus, results in higher observed phosphorescence. 

The DTPA impregnated S&S 903 substrate demonstrates superior 
phosphorescence at room temperature because of a different mechanism. 
It was initially thought that the addition of DTPA would remove trace 
amounts of transition metals in filter paper substrates (164) that might 
contribute to phosphorescence background. Treatment with DTPA, however, 
gave no improvement in blank signals, but did significantly improve the 
analyte signal. The effect on the analyte signal was attributed to 
interaction of the analyte with DTPA (a strong hydrogen bonding agent) 
and to filling of the gaps in porous filter paper substrates. This 
resulted in analyte molecules being trapped in the DTPA-cellulose matrix 
(142) and causing higher phosphorescence intensities as a result of 
stronger binding to the substrate and a higher surface concentration of 
analyte molecules because of less penetration of the analyte into the 
substrate. Another possible explanation of the improved analyte signal 



155 



when DTPA is added to the S&S 903 substrate is that metal ions whose 
interaction with the analyte results in decreased phosphorescence are 
chelated. In this way, the analyte would be protected from interaction 
with these metal ions and higher phosphorescence intensities would 
result. 

The faster drying time observed for the BTPA treated S&S 903 
substrate could also be due to the higher surface concentration of 
analyte molecules. The flow of dry nitrogen is directed onto the 
surface of substrate sample discs, therefore, the water removing 
nitrogen would act faster on molecules concentrated on the surface of 
the substrate than for those which have penetrated into the depths of 
the paper. The DE-81 filter paper is also considerably denser than the 
DTPA treated S&S 903 filter paper and water removal from the denser 
substrate is expected to take a longer time. The rise times of 
phosphorescence were shortened and the plateau times were increased for 
both substrates when 50% ethar.ol was used as the spotting solvent. The 
increase in phosphorescence rise time observed when ethanol was added 
was probably due to increased volatility of the ethanol-water mixture as 
compared to water alone. Plateau times may be related to the strength 
of analyte binding to the substrate, since decreases in phosphorescence 
following the plateau are probably due to desorption of analyte 
molecules. The BTPA treated S&S 903 substrate demonstrated longer 
plateau times and slower fall rates which suggests that its interaction 
with analyte molecules is probably stronger than the ionic interaction 
of the analyte with charged groups on DE-81. This is also suggested by 
the observation that DTPA treated S&S 903 demonstrated greater 
phosphorescence signals than DE-81 without the heavy atom added. The 



156 



fact that the addition of 50% ethanol increased the plateau times on 
both substrates is probably related to the lower viscosity of this 
medium which causes greater spreading of analyte molecules along with 
greater penetration into the substrate. A corresponding decrease in 
total phosphorescence intensities with 50% ethanol added to the 
application mixture was not observed, however, which does not support 
this claim. The decrease in fall rate for DTPA treated S<SS 903 observed 
when ethanol was present in the application mixture also does not 
support this claim and an exact explanation of the mechanistic effect of 
ethanol is not possible with the data presented here. 

The environment of the DE-81 substrate is basic (151) and the 
diethylaminoethyl group may be expected to have some buffer capacity. 
This explains the extended pH range in which the DE-81 substrate 
provided relatively stable phosphorescence intensities. The range is 
extended to a much greater extent in the basic region for DE-81 as 
compared to the DTPA impregnated S&S 903 substrate. This is because of 
the basic nature of the DE-81 substrate and its ability to buffer 
solutions applied to it in the basic pK range. The range of stable 
phosphorescence in the acidic region is also slightly extended for the 
DE-81 substrate, suggesting that the substrate has the ability to 
neutralize acidic spotting solutions to some extent. The 
phosphorescence intensity versus pH curve for the DTPA treated S&S 903 
substrate was stable over a smaller range of pH because of an apparent 
lack of buffer capacity of the substrate. 

The decrease in phosphorescence signal in strongly acidic solutions 
(below pH 2.5) is probably due to protonation of either the carboxyl 
group or the amino group on PABA which prevents strong binding to the 



157 



substrate. The pH at which this decrease occurs for both substrates 
supports protonation of the amino group as the primary cause of the 
decrease and the fact that the quaternary ammonion group on DE-81 would 
not be expected to bind strongly with a positively charged analyte also 
supports this. Von Wandruska and Hurtubies (66) proposed that the 
decrease observed at high pH in the room temperature phosphorescence of 
PABA on a sodium acetate substrate was due to precipitation of NaOH upon 
evaporation, but did not offer an explanation why this would result in 
lower phosphorescence intensity. A possible explanation for the loss of 
phosphorescence intensity in the presence of precipitated NaOH is the 
hygroscopic nature of NaOH crystals. The water attracted to the 
substrate by NaOH crystals would result in lower RTP intensities (65). 
This study suggests that there may be a competition for substrate 
binding sites because the DE-81 substrate has a much greater anion 
binding capacity than the DTPA treated S<5S 903 and also showed stronger 
RTP in the basic region. Another possible contribution to the decrease 
in RTP intensities observed in the basic region for DTPA treated SfiS 903 
is the possible loss in chelation of metal ions, which interact with the 
analyte, by DTPA in that pH range. 

The higher RTP intensities observed from the DE-81 substrate in the 
presence of iodide is easily explained in terms of heavy atom binding to 
the substrate. The anionic exchange filter paper, DE-81, binds anions 
such as iodide making them more accessible to analyte ions which are 
also bound to the substrate. 

The increased iodide heavy atom effect for DE-81 as compared to 
DTPA treated S&S 903 was present in both blank and analyte studies which 
suggests that the observed effect is a property of the substrate and not 



158 



the analyte-heavy atom interaction on the substrate. The positively 
charged heavy atom species tested on the DE-81 substrate gave lower room 
temperature phosphorescence intensities than untreated DE-81 which 
suggests either competition of the analyte with heavy atom species for 
substrate binding sites or an increased rate of radiationless 
deactivation processes caused by the spin-orbital coupling effect of the 
heavy atom (165). 

The performance of the RTP method in clinical samples was observed 
to have certain advantages and disadvantages when compared to other 
analytical methods. The range of linearity for the RTP system is 
superior to that of both colorimetric methods (21,42) but not as wide as 
that for HPLC with electrochemical detection (154). The RTP method is 
less precise than both colorimetric and HPLC procedures, but analytical 
variability is overshadowed by physiological variability which is 
encountered in the bentiromide test and the contribution to method error 
due to analytical imprecision is small for the RTP method. 

Another advantage to RTP analysis by the proposed method is 
complete conversion of all PASA metabolites back to the parent 
compound. The acid hydrolysis, which is commonly used in colorimetric 
procedures, does not completely convert PAAHA to PABA but rather 
converts it to the intermediate PAHA. Fortuitously, this metabolite 
reacts in both colorimetric procedures to form a chromophore similar to 
that of the PABA-chromogen complex. However, any differences in 
reactivity or absorptivity of these complexes combined with the observed 
individual variability in PABA metabolism could lead to erratic 
results. Additionally, colorimetric methods are non-specific and 
urinary aromatic amine contamination would cause false positive 



159 



results. Extreme care must be taken to avoid these interferences and 
interfering drugs must be discontinued at least 3 days prior to the 
test. This might necessitate the interruption of critical drug therapy 
which is not possible, in some cases. Electrochemically detected HPLC 
is seemingly not subject to such interferences although complete drug 
interference studies have not been carried out with the HPLC method. 

The RTP method is relatively specific with regard to the compounds 
tested except for procaine which does not pose a therapeutic problem if 
discontinued. Also, metabolite recovery studies and sample blank 
analysis indicate that endogeneous urinary components and hydrolysis by- 
products do not interfere with the method. The detection limit is more 
than adequate for clinical analysis of urine samples and provides the 
potential for PABA analysis in blood. 

The RTP method is more rapid and convenient than either KPLC or 
colorimetric procedures and readily applicable to routine analysis. 
Although some of the equipment necessary is not commercially available, 
slight modifications of commercially available front surface attachments 
could be used with slight modification C 1 41 ) - The diagnostic 
performance of the bentironide test with RTP detection of recovered PABA 
can be considered to be essentially equivalent to that for colorimetric 
methods if care is taken to avoid drug interferences. This is validated 
by the good correlation of the two methods under these conditions and 
the diagnostic sensitivity and specificity for the method would be 
roughly equivalent to 12% and 95£, respectively. 

Variables Involved in the Chromatographic Technique 
Brief explanations regarding the observations that were made during 
development of the HPLC procedure for PABA and its metabolites have 



160 



already been presented in Chapter 3 of this dissertation to rationalize 
the choices made. Therefore, a discussion of these factors here will be 
limited to an overview of the factors involved and the analytical merits 
of the system. 

It was suggested in the results section of this work that methanol 
was a less selective solvent (produced smaller k' values) for adjusting 
the capacity factors of PAABA and PAAHA as compared to acetonitrile. 
The proposed explanation is that the amides ( PAABA and PAAHA) are better 
hydrogen bonding species (154) than the amines (PABA and PAHA) and 
would, therefore, interact more strongly with a proton donating and 
accepting solvent (155) such as methanol. The more strongly the solvent 
interacts with specified sample components, the more preferentially it 
will dissolve the sample compound in question, thus decreasing the 
retention provided by the stationary phase. There are actually three 
major interactions between solute and solvent molecules that must be 
considered in determining the primary mechanism of retention; 
dispersion, dipole and hydrogen bonding forces. 

Dispersion interactions which result from induced dipolar moments 
are stronger for compounds having highly polarizable electrons and are 
related to the refractive indices of solvent and solute molecules. The 
refractive indices of acetonitrile and methanol are 1.341 and 1.326 
(156), respectively, which suggests a very small difference in the 
strength of dispersion interactions between the two solvents. This 
indicates that the differences in retention observed for analyte 
molecules in methanol and acetonitrile are probably not due to 
dispersion interactions. 



161 



Dipole interactions are observed to occur between molecules having 
permanent dipole moments. These interactions occur with solvent and 
sample molecules that have relatively polar functional groups and the 
interaction is stronger with functional groups that are strongly 
electron donating or withdrawing. The dipole moments of the hydroxyl 
group on methanol and the nitrile group on acetonitrile are 1.7 and J. 5 
debyes, respectively (110). This relatively large difference in dipole 
moments between the main functional groups of the two solvents studied 
indicates that there is a significant difference in dipole interactions 
with solute molecules. The interaction should be stronger for 
acetonitrile, however, and this is not consistent with the observations 
made in this work. This suggests that differences in dipole 
interactions were not the primary reason for the preferential 
interaction of solvent molecules with PAABA and PAAHA. 

Hydrogen bonding interactions were chosen as the likely mechanism 
to explain greater interactions between methanol and the acetylated 
analyte molecules because of the greater hydrogen bonding capacity of 
the amide group (155). Amides represented by PAAEA and PAAHA interact 
more strongly with hydrogen bonding solvents by virtue of their ability 
to form multiple hydrogen bonds. Although the amino substituents on 
PABA and PAHA are more strongly basic and thus better hydrogen bond 
acceptors, the amido substituents can form hydrogen bonds at both the 
nitrogen and carboxyl oxygen sites. This creates a combined effect for 
the amido compounds that is stronger than the individual effect 
corresponding to the amino derivatives. 

The concept of ion-pair equilibria which was introduced in Chapter 
1 and referred to in Chapter 3 requires further discussion here so that 



162 



the results observed may be described in more detailed terms. The 
following set of simultaneous equilibria is necessary to adequately 
describe the entire system of ion-pair chromatography: 



[rcoct] 



m 



k-1 



[rcoo~], 



IP^ 



IP^ 



(30) 



k-3 



k-2 



k-4 



IP + [RC00"j r 



k-5 



IP + [RC00"j ( 



where RCOO" represents negatively charged analyte molecules and IP + 
represents a positively charged pairing-ion molecule. The subscripts m 
and s signify mobile and stationary phases, respectively, and the 
foreward and reverse rate constants are represented by k with positive 
or negative subscript numbers indicating forward and reverse rates, 
respectively. The simultaneous equilibria described in (30) shows * 
distribution of free analyte ions, pairing-ions and ion pairs between 
the mobile and stationary phases at rates corresponding to k 1 , k 2 and 
k^. The diagram also shows that ion-pairs are formed at different rates 
in the mobile and stationary phases. 

It is very difficult to determine the exact cause for observations 
involving ion- pairing agents with only capacity factor data, because the 
observed effects could be due to a number of contributing factors. For 
example, it was proposed in the results section of this work that the 
deviation from linearity observed for the pairing agent hexadecyl- 
trimethylanmonium bromide, which did not belong to the homologous series 



163 



represented by the other pairing agents, was due to either alteration of 
the stationary phase because of pairing-ion binding or an increased ion- 
pair formation constant because of better accessibility of the 
hexadecyltrimethylammonium charged head. It is not possible to 
distinguish the cause of nonlinearity with capacity factor data and each 
possible cause could in turn be due to several factors. Alteration of 
the stationary phase could be due to increased hydrophobicity attributed 
to binding of the pairing ion (k 2 ) or an apparent ion exchange effect 
represented by the rates k 1 and k 4 . The higher potential accessibility 
of the charged head on hexadecyltrimethylammonium bromide could also be 
due to either a higher ion-pair formation rate in the mobile- phase (k-.-,) 

J 

or a higher ion pair formation rate in the stationary phase (k,). It is 
also obvious from the equations in (30) that changing any of the rates 
will in turn affect the other rates and further complicate definitive 
evaluation. 

In the study of anionic ion-pairing agents, it was observed that 
the capacity factors of PAAEA and PAAHA increased at pH 2.5 with 
increasing pairing-ion size. The pKa's of the amido groups are 
considerably lower than 2.5 (see Figure 3-12), however, which suggests 
that retention by the pairing-ion must involve alteration of the 
stationary phase because ion-pair formation with these compounds would 
not be expected at that pH. The overall effect on retention by anionic 
pairing agents was much greater for PAHA and PABA which indicates ion- 
pair formation must be considered because these analytes are ionized at 
the pH studied. These two observations are good evidence for the 
combined effect which was proposed in the results section, but the exact 
contributions cannot be determined. 



164 



The HPLC method for PABA and its metabolites that was developed is 

only slightly less precise than the method of Ito (153) for analysis of 

PABA in which an internal standard was used. The HPLC method was proven 

useful for differentiation of metabolite excretion in liver and 

pancreatic patients if the test conditions are carefully controlled, 

which reflects good accuracy of the method in clinical samples. With 

modifications, the method can be applied to the quantitation of PABA and 

its metabolites in blood to enable in depth pharmacokinetic studies that 

may reveal a better understanding of PABA metabolism. 

Detection or Falsely Positive Bentiromide Test Results 
by Ketabolite Analysis 

Although the primary pathway of PABA metabolism cannot be 

definitively obtained by analysis of metabolites in urine samples, there 

was some information gained in this study which can be treated in 

qualitative terms. The pathways that could have resulted in the 

metabolites that were found are shown in (31). 



PABA * PAABA * PAABG 

PAHA * PAAHA (31) 

PAHG 



Para-acetamidobenzoyl glucuronide and p-aninohippuryl glucuronide are 
represented in (2) by PAABG and PAHG, respectively. These two compounds 
were shown to exist in urine samples of subjects undergoing the 
bentiromide test by recovery of PAHA and PAABA after treatment with 
glucuronidase. The observation that PAABG was only found in urine 
samples collected at least 2 hours after bentiromide administration and 
after high levels of PAABA were reached, indicates that PAABG is formed 



165 



from PAABA and not through a glucuronide conjugated intermediate of 
PABA. All three groups of subjects studied demonstrated a much greater 
ratio of PAABA to PAAHA in earlier (0-3 hr collection interval) samples 
than in later samples (3-6 hr collection interval) which indicates that 
conversion of PAABA to PAAHA is the primary pathway for PAAHA synthesis 
rather than PAHA to PAAHA. This is also substantiated by the fact that 
no appreciable amounts of PAHA were found in urine samples and that the 
amount of PAHG was also a small fraction of the total amount of 
metabolites. The amount of PAAHA produced in normal subjects is 
approximately equal to the amount of PAABG as shown by HPLC analysis of 
PABA and its metabolites, in samples before and after treatment with 
glucuronidase. This suggests that the rates of formation of PAABG and 
PAAHA from PAABA are similar. 

All of these interpretations should be regarded as only qualitative 
descriptions of the most likely metabolic pathway since good 
quantitative pharmacokinetic data could not be obtained from the urine 
samples. It can be said with certainty that the major metabolites of 
PABA in all groups of clinical subjects studied are PAABA, PAAHA and 
PAABG. These results agree with studies done in dogs (27) and guinea 
pigs (28) where two dimensional thin layer chromatography and liquid- 
liquid extraction were used as the separation techniques. Following 
treatment with glucuronidase, the major metabolites present for the 
differentiation of metabolism in liver patients are PAABA and PAAHA. It 
was shown in the results section of this work that the most significant 
parameter for the evaluation of liver function by PABA netabolism was 
the ratio of PAABA to total PAABA + PAAHA in glucuronidase treated urine 
samples collected over a 6 hour collection interval. The results from 



166 



HPLC analysis of PABA metabolites were shown to be more reliable for 
this purpose than analysis by differential hydrolysis of PAABA and PAAHA 
followed by RTP measurement. 

The 95% confidence interval for mean ratios from pancreatic and 
liver population is 0.363-0.508 and 0.480-0.813, respectively, as 
analyzed by HPLC. Although the means of these populations were shown to 
be significantly different at p<0.01, there is some overlap of the two 
95%' confidence intervals which is more closely related to the clinical 
utility of the method. The 95% confidence interval can be used as a 
value for the normal range within each population that would include 95%' 
of that population. Defined in those terms, there is a 19.0% overlap of 
the pancreatic population with that of the liver population and an 3.4% 
overlap of the liver population with the pancreatic population. These 
Figures can be related to diagnostic sensitivity and specificity because 
the number of false positives (pancreatic patients diagnosed as having 
liver dysfunction) is related to the degree of overlap between the 
pancreatic patient confidence interval and the liver patient confidence 
interval. Similarly, the degree of overlap of the liver patient 
population with that of the pancreatic patient population is related to 
the number of liver patients that would be evaluated as not having liver 
dysfunction (false negatives). The value for diagnostic sensitivity can 
then be predicted to be 84.5% and the diagnostic specificity 68.1% which 
reflects the percent of positive tests that were accurately tested as 
positive and the percent of negative tests that were accurately tested 
as negative, respectively. If the cut-off point for liver dysfunction 
is taken to be the lower limit of the 95% confidence interval for the 
mean ratio of PAA2A to PAABA + PAAHA, only 2 



167 



out of the 10 pancreatic patients tested in the clinical study would 
have had ratios above the cut-off point and been tested as positive for 
liver dysfunction. This yields an observed diagnostic specificity of 
66$ which agrees well with the predicted value. Among the 5 liver 
patients tested, none had ratios that were below the cut-off point so 
the observed diagnostic sensitivity was 100%'. This does not agree well 
with the predicted value although better agreement would probably be 
obtained with a larger sample of the liver patient population. 

The diagnostic value of PAABA to PAABA + PAAHA ratios determined by 
differential hydrolysis with detection by RTP would not be expected to 
be as good because of the observed interferences from unconverted 
metabolites in the acid hydrolysis procedure. The 95% confidence 
interval for liver and pancreatic patient populations in this case were 
0.625-0.962 and 0.519-0.735, respectively. The degree of confidence 
interval overlap for the liver population was 12.6% and that for the 
pancreatic population was 50.9$. The predicted values for diagnostic 
sensitivity and selectivity with the RTP method were 50.8% and 32.5%, 
respectively. The observed diagnostic sensitivity in the clinical 
samples was 42.8% and the diagnostic selectivity was also 42.8%. It is 
obvious from these calculations that the diagnostic value of the RTP 
method for metabolite analysis is inferior to that of the HPLC method. 
Although the RTP method is more selective analytically with regard to 
endogeneous components of urine, the poor diagnostic value of the 
results makes the method less useful for the defined purpose. 



165 



Physical Aspects of the Room Temperature Phosphorescence 
of Para Aminobenzoic Acid and Its Metabolites 

Various studies related to the phosphorescence characteristics of 

PABA and its metabolites were carried out in an effort to resolve their 

RTP signals for analytical purposes. Although these attempts were 

unsuccessful because of inadequate differences among the wavelengths of 

maximum emission, heavy atom effects and lifetimes of the individual 

analytes, the data allow a discussion of the physical aspects of these 

observations which will be presented here. Before a detailed discussion 

of the observed RTP characteristics can be presented, it is necessary to 

briefly describe origins of the lowest excited triplet state. 

The phosphorescent properties cf a molecule are determined by the 

nature of population of the lowest excited triplet state. The rates of 

intersystem crossing (ISC) from singlet (S, ) to triplet (T, ) states 

generally increase as the energy difference between these states 

decrease (166). Singlet and tr olet states from tttt* electronic 

configurations that are derived from transitions involving lone pair 

electrons on substituent groups lie lower in energy than inr* states of 

the unsubstituted compound. This will result in the substituted 

molecule having a smaller degree of singlet-triplet energy splitting, if 

the lowest excited triplet rrn* state is less shifted to lower energy 

than the lowest excited singlet tttt* state. The smaller degree of energy 

splitting in the substituted molecule would result in an enhanced rate 

of ISC and contribute to enhanced phosphorescence efficiency. 

Benzoic acid did not show any phosphorescence at room temperature 

whereas the amino derivative of benzoic acid (PABA) showed very strong 

phosphorescence. This observation could be due to the S, and T, states 



169 



in PABA originating from electronic configurations involving lone pair 
electrons on the amino nitrogen that are closer in energy than those of 
benzoic acid. The belief that S 1 is lower in PABA is substantiated by a 
34 nm red shift in the absorption spectrum of PABA relative to benzoic 
acid. 

The metabolic conjugation of the substituent groups on PABA results 
in derealization of the electron donating and accepting effects 
attributed to these substituents. This is one possible explanation for 
the lower phosphorescence intensities observed for metabolites of 
PABA. The phosphorescence intensities of both the acetyl derivative 
(PAABA) and the glycine derivative (PAHA) are about 14- fold lower than 
that of PABA. When PABA is both acetylated and glycine conjugated 
(PAAHA) the phosphorescence at room temperature is decreased approxi- 
mately 133- fold. The phosphorescence emission maximum of PAHA and PAAHA 
were red-shifted 10 and 9 nm, respectively (relative to PABA ) whereas 
the PAABA conjugate is blue-shifted only 3 nr.. The corresponding 
excitation maxima were red-shifted 6 nm for PAHA and 1 nm for PAAHA and 
blue-shifted 17 nm for PAABA. The resulting energy splitting of the S 1 
and T 1 states are 34 nm for PABA, 33 nm for PAHA, 48 nm for PAABA and 42 
nm for PAAHA. The greater energy splitting of S 1 and T 1 for all 
metabolites of PABA relative to the parent compound is a possible 
explanation for the decrease in phosphorescence efficiency of these 
compounds. This is a very simplified explanation, however, because both 
the radiative and nonradiative processes that deactivate the triplet 
state depend on not only S 1 and Ti but also on the nature of the 
intermediate triplet states which affect the deactivation process. 
Therefore, alterations in the degree of energy splitting between higher 



170 



triplet states and T 1 , which may have been caused by the delocalizing 
effect of metabolic conjugation on PABA substituents , could have 
resulted in lower phosphorescence intensities. 
Lifetimes and Heavy Atom Effects 

The presence of heavy atoms of high atomic number in the vicinity 
of a phosphorescent molecule enhances its probability of phosphorescence 
by increasing the rate of intersystem crossing and the rate of phos- 
phorescence from the triplet state. The heavy atom effect can affect 
both radiative and nonradiative singlet-triplet transitions and is 
purely electronic in nature ( 1 67 ) . The quantum yield of phosphorescence 
(0p) is defined by 

k p 



p " &>T k p + k TS (32) 

where k p is the rate constant of phosphorescence, k TS is the sum of all 
unmolecular or pseudo first order rate constants for radiationless 
deactivation of the triplet state and ST is the efficiency of singlet- 
triplet crossing from the lowest excited singlet state. Equation (32) 
shows that phosphorescence quantum yields can be decreased by the heavy 
atom effect if krp S is increased to a great extent relative to 0gm and 
kp. This is a possible explanation for the decrease in phosphorescence 
intensity observed for PABA in the presence of Tl + and Ag + . However, 
the phosphorescences of all of the metabolites of PABA were enhanced by 
Tl and Ag + which suggests that in these compounds k„ and/or 0™ 

p Oi 

increase faster than k TS as a result of enhanced spin-orbital coupling. 

The relative effects of the iodide heavy atom on phosphorescence 
intensities and lifetimes of PABA and its metabolites demonstrate a 



171 



pattern which can be related to the type of metabolite conjugation. In 
order to discuss this pattern, the lifetime of the triplet state ( T ) 
must be defined in terms of intersystem crossing and triplet state 
deactivation rates as follows. 

0ST 

T P = k p + k TS (33) 

In qualitative terms, if Tp is increased ST increases faster than k p + 
k Ts . Such is the case with all PABA metabolites relative to PABA and 
the glycine conjugates demonstrate the greatest increase in T . 
Correspondingly if p is decreased, as is also the case with the 
metabolites of PABA, s , r k p decrease faster than k p + k TS (e quation 
32). Therefore, k p decreases faster than k TS for all metabolite 
molecules relative to PABA since ST in the metabolites is increased 
relative to k p + k Tg . This suggests that the increase in 
phosphorescence lifetimes observed for the metabolites of PABA can be 
attributed more to a decrease in k p rather than a decrease in k T g. 
Also, the phosphorescence quantum yields of PABA metabolites were 
decreased by decreases in k p since ST /k p + k ?s is increased for all 
compounds. The portion of phosphorescence intensity decrease which was 
due to decreases in k p is greater for PAHA and PAAHA than for PAABA 
because of the longer lifetimes of the former. The amount of 
phosphorescence intensity decrease relative to PABA was about the same 
for PAHA and PAABA so the phosphorescence intensity decrease due to 
increases in kng and or decreases in ST must have been greater for the 
acetylated derivative. 



172 



This can be shown in quantitative terms by dividing equation 32 by- 
equation 33 to yield the following expression. 

i 

T p ■ * p (54) 

Although the absolute quantum yield of room temperature phosphorescence 

was not determined because of the lack of a quantum yield standard, the 

relative k values can be calculated for each compound of interest by 

determining phosphorescence intensities at concentrations that would 

yield the same absorbances and substituting these values for quantum 

yields. The values of k p calculated in this manner were 0.93 for PABA, 

0.12 for PAHA, 0.18 for PAABA and 1.2x10" 5 for PA AHA. 

The relative contributions of gT and or k TS can be estimated by 

solving equation (33) for 0c; T and substituting values for k and T . 

■~ J - P P 

The equation for each compound can then be normalized so that it is in 
the form: 

X ST - Y k TS = 1 (35) 

where X and Y are coefficients derived from the data for k p and vhich 
are inversely related to the relative magnitudes of 0™ and k mo . 
Although the values for ST and kmg can not be separated by this 
technique, the sum of contributions from r .r n + kmc- can be ordered from 
highest to lowest showing relative effects. The order of contributions 
from gT + k Tg for the compounds of interest were found to be 
PABA > PAABA > PAHA > PAAHA. 

The values calculated for k p show that the glycine conjugated 
derivative of PABA loses phosphorescence from a decrease in k p upon 



173 



conjugation more so than does PAABA. The order of 0™ + k™ indicates 

Ol lo 

that PAABA loses phosphorescence because of a decrease in ST and or an 
increase in k ST more so than does PAKA. The PAAKA metabolite shows the 
greatest decreases in phosphorescence by both mechanisms which repre- 
sents the combined effect of altering both substituents on PABA. 

In the presence of the iodide heavy atom, the relative values of k 

P 

were calculated to be 7.3 for PABA, 2.2 for PAHA, 2.0 for PAABA and 0.32 
for PAAHA. The interesting aspect here is that the relative magnitudes 
of the k p values for PAHA and PAABA are reversed upon addition of the 
heavy atom. The order of the degree of contributions from 0™ + k mo for 
PABA and its metabolites is PABA > PAHA > PAABA > PAAHA. Again, the 
relative contributions to a loss of phosphorescence intensity by 
decreases in ST and or increases in k TS were reversed for PAHA and 
PAABA. There were no changes observed in the energy splitting of the 
lowest excited triplet state and the ground singlet state in the 
presence of the heavy atom. 

There- are several statements that can be made with certainty about 
the relative phosphorescence characteristics of PABA and its metabo- 
lites. The rate constant of phosphorescence is decreased by both acety- 
lation and glycination of PABA. The combined decreases in rates of 
singlet-triplet intersystem crossing and nonradiative deactivation of the 
triplet state which contribute to a loss of phosphorescence intensity are 
enhanced by both acetylation and glycination. The combined effect cf 
acetylation and glycination decreases phosphorescence intensity to a 
degree that is greater than the sum of the individual effects. There is 
also a difference in the mechanism of phosphorescence reduction between 



174 



glycination and acetylation which is reversed upon treatment with the 
iodide heavy atom. 

Exactly how these observations relate to the origin of electronic 
states and the delocalizing effects of metabolic conjugation can only be 
speculated. It is reasonable to assume, however, that the effects ob- 
served for acetylation of the amino group are related to interaction of 
the electron withdrawing acetyl group with the nit- state originating from 
transitions involving lone pair electrons on the amino group. Also, the 
effects observed for glycination must be due to a direct effect on the irir* 
state originating from transitions involving the transfer of electron 
density from the aromatic ring to the carboxyl group. The trir* states that 
arise from charge transfer from the aromatic ring to the carboxyl group 
should be viewed as involving charge transfer from the lone pair of the 
donor (amino group) to the v orbital associated with the acceptor 
(carboxyl group). It appears from the data that glycination of the 
carboxyl group results in an increase in its electron withdrawing ability, 
thus decreasing the energy of transition. This was observed as red shifts 
in both the excitation and emission phosphorescence maxima for PAHA. The 
blue shifts associated with acetylation of the amino group are probably 
due to a decreased electron donating effect since the acetyl group is more 
strongly electron withdrawing than the hydrogen atom. 

In all cases, these shifts in tttt* state energies resulted in an 
increased energy gap between S, and T 1 relative to PABA. The observation 
that the decrease in phosphorescence of PA ABA was due more to decreases in 
ST and/or increases in k TS rather than decreases in k p (relative to PAHA) 
is consistent with the observed spectral shifts since the energy gap 
between S 1 and T 1 is larger for PA ABA . 



175 



The Effect of Substrate Binding 

The effect of analyte binding to the substrate must also be con- 
sidered as a source of the observed differences in phosphorescence 
intensities. The amino and carboxyl groups on PABA can interact with the 
DE-81 substrate by hydrogen bonding of the amino groups to hydroxyl groups 
on the substrate and by ionic interactions of the carboxylate group with 
the alkyl ammonium ion of the substrate. A similar mechanism for the 
binding of PABA to sodium acetate was described by von V.'andruszka and 
Hurtubise (66), which showed that this type of parallel orientation of 
analyte molecules to the solid surface maximized binding. 

The lower phosphorescence intensities observed for glycinated and 
acetylated derivatives of PABA could be due to weaker interactions with 
the substrate because of the derealization of charge on the substituent 
groups. The carboxamido group on PAHA may be less conjugated with the 
aromatic ring than the carboxylate group on PABA and less able to withdraw 
electron density from the ring. This would result in an apparent decrease 
in the negative charge of the carboxamido group on PAHA relative to the 
carboxylate group of PABA and may result in a weaker interaction. The 
acetylated amino group on PAABA could also cause a change in the charge 
density relative to PABA. This may result in weaker binding because the 
amido group on PAABA is less basic than the amino group of PABA. The 
acetylated conjugate, however, also has a carboxyl oxygen atom that could 
contribute to hydrogen bonding with hydroxyl groups, thus enhancing 
binding. 

Another possible mechanism for weaker binding of the glycine 
conjugates is that of intramolecular hydrogen bonding between amino and 
carboxyl groups. This would compete with intermolecular hydrogen bonding 



176 



to the substrate and would result in weaker bonding. The relative binding 
of PABA and its metabolites cannot be determined without further study and 
the mechanisms described here should be considered hypothetical. 

Conclusions 

This study on the analytical aspects of the bentiromide test has 
yielded data which indicate that room temperature phosphorimetry is a 
superior analytical method to those in current use. Analysis of 
bentiromide metabolites by ion-pair high performance liquid chromatography 
has shown that ratios of certain metabolite concentrations can be used to 
differentiate false positive bentiromide test results due to liver 
dysfunction. These metabolite concentration ratios can also be determined 
by room temperature phosphorimetry with differential hydrolysis although 
this method is inferior to the chromatographic method in terms of 
diagnostic value. The appropriate urine collection interval for analyses 
of metabolite ratios was determined to be six hours and the major 
metabolites of bentiromide were identified as p-acetamidobenzoic acid, p- 
acedamindohippuric acid and p-aceamidobenzoyl glucuronide. 

In addition to analytical and clinical studies, the physical aspects 
of the phosphorescence of bentiromide metabolites were studied. The rate 
constant of phosphorescence at room temperature is decreased to a 
different degree by both acetylation and glycine conjugation of p- 
aminobenzoic acid. The singlet-triplet intersystem crossing efficiency is 
decreased and/or the rate constant for non- radiative deactivation of the 
excited triplet state is increased by both acetylation and glycination of 
p-aminobenzoic acid. The combined effects of acetylation and glycination 
decrease phosphorescence at room temperature to a degree that is greater 
than the sum of the additive individual effects. 



APPENDIX I 
INTERFERING SUBSTANCES 



acetaminophen 

acetophenetidin 

aspirin 

atropine 

benzocaine 

chloramphenicol 

chlorothiazide 

indomethacin 

lidocaine 

metaclopramide 

neomycin 

oral pancreatic enzyme supplements 

phenoba rbital 

procaine 

procainamide 

sulfonamides 

thiazide diuretics 



177 



APPENDIX II 
CLINICAL EVALUATION 

1. History - to include biographic data, history, and review of 
systems. 

2. Physical Exam - to include height, weight and a thorough systematic 
examination. 

J>. Vital Signs - to include supine blood pressure, pulse, respiration 
and body temperature (oral). 

4. Clinical laboratory - to include the following: 

Hematology - hematocrit, red blood cell count, total white blood 

cell count and differential white blood cell count. 
Serum chemistry - BUN, SGOT, alkaline phosphatase and creatinine. 
Urinalysis - specific gravity and urine qualitative chemistry. 



178 



APPENDIX III 
EXCLUSION CRITERIA 

1 . History of having received any investigational drug during the four 
weeks prior to entrance into study. 

2. History of hepatic, pancreatic or renal disease. 

3. History of alcoholism or excessive alcoholic consumption. 

4. History of malabsorption. 

5. Treatment with any drugs listed in Appendix 1 at time of initial 
evaluation. 

6. Presence of any significant pretreatment clinical laboratory 
findings. 

7. Requirement for any concomitant medication listed in Appendix 1 
during the period of the study. 



179 



REFERENCES 

1. T. Scratcherd, Gut, 16 (1975) 648. 

2. C. Arvanitakis and A. R. Cooke, Gastroenterology, 74 (1978) 932. 

3- C. Arvanitakis, P. P. Toskes and N. Greenberger, 

Gastroenterology, 74 (1978) 1004. 

4. E. P. Di Magno, V. L. W. Go and W. H. J. Summerskill , N. Engl. J. 
Med., 288 (1973) 813- 

5. M. I. Grossman, Gastroenterology, 60 (1971 ) 157. 

6. W. B. Salt, II, and S. Schenker, Medicine (Baltimore), 55 (1976) 
269. 

7. G. Lundh, Gastroenterology, 42 (1962) 275. 

8. E. P. Di Maquo, V. L. W. Go and W. H. J. Summerskill, 
Gastroenterology, 63 (1972) 25. 

9. A. R. Imondi, R. P. Stradley and R. Wolgemuth, Gut, 13 ( 1 97 1 ) 
726. 

10. A. R. Imondi, R. P. Stradley, R. Wolgemuth and T. G. Brown, 
Pharmacologist, 13 (1971 ) 290, Abstract. 

11. K. Gyr, R. H. Wolf and A. R. Imondi, Gastroenterology, 68 (1975) 
488. 

12. R. Gyr, G. A. Stadder, J. Schiffman, C. Fehr, D. Vonderschmett 
and H. Fahrlander, Gut, 17 (1976) 27. 

13. S. Nousia-Arvanitakis, C. Arvanitakis and N. D. Greenberger, J. 
Pediatrics, 92 (1978) 734. 

14. T. Kimura, H. Wakasugi and H. Ibayashi, Digestion, 21 (1981 ) 133. 

15. C. Lang, K. Gyr and G. A. Stadler, Brit. J. Surg., 68 (1981 ) 771. 

16. W. Bornschien, F. L. Goldman and J. Dressier, Klin. Wocheschr. , 
56 (1978) 197. 

17. C. J. Mitchell, C. S. Humphrey and A. W. Bullen, Scand. J. 
Gastroenterol., 14 (1979) 183. 



180 



131 



K. Gyr., "Pancreatic Function Draquostant" , Ed. by M. Kasoda, 
Iaaku-Shoin, New York, 1980, p. 38. 

P. P. Toskes, Gastroenterology, 85 (1933) 565. 

J. G. Salway and R. B. Payne, Lancet, 2 (l9 r ?6) 100. 

H. V. Smith, H. Frankelstein, L. Aliminosa, B. Crawford and M. 
Graber, J. Clin. Invest., 24 (1945) 368. 

C. J. Mitchell, C. S. Humphrey and A. W. Bullen, Scand. J. 
Gastroenterol., 14 (1978) 183. 

C. J. Mitchell, C. S. Humphrey, A. W. Bullen, J. Kelleher and M. 
S. Losowsky, Scand. J. Gastroent., 14 (1979) 737. 

P. P. Toskes, Pharmacotherapy, 4 (1934) 74. 

P. P. Toskes and N. J. Greenberger, P. M. , 29 (1983) 1. 

R. Stradley, W. Rogers, R. Sheding, California Veterinarian, o 
(1981) 29. 



C. Yamato and K. Kinoshita, J. Pharmacol. Exp. Ther., 206 (loip.) 
468. 

*•!. Kitamura and M. Nakao, J. Biochem. , 47 (i960) 60. 

W. B. Jakoby, "Enzymatic Basis of Detoxification Vol. II", 
Academic Press, London, 1980, p. 5. 

W. P. Deiss and P. P. Cohen, J. Clin. Invest., 29 (1950) 1014. 

C. W. Tabor, J. Bailey and P. K. Smith, Fed. Proc, 6 (1947) 376. 

M. M. Drucker, S. II. Blondheim and L. V/islicki, Clin. Sci. 27 
(1964) 133. 

J. C. Delchies and J. C. Soule, Gut, 24 (1 Q 83) 313. 

F. J. Hoek, G. T. E. Saunders and A. Teunen, Gut, 22 (1981) 6. 

G. H. Kay, V. A. Tetlow and J. J. Braganea, Clin. Chem. Acta., 
128 (1983) 115. 

J. Slaff, C. Currington and C. King, Castroenterolo,£?v. 64 (1983) 
1314. 

E. E. Sterchi, J. R. Green and M. J. Lentze, Clin. Sci.. 62 
(1982) 557. 

C. S. King and ?. P. Toskes, Gastroenterology, 76 (1979) 1035. 



182 



39. I. Shosoki, K. Maruta, Y. Imai, T. Kato, M. Ito, S. Nakajima, K. 
Fujita and T. Kurahasi, Clin. Chem. , 28 (1982) 323. 

40. C. Arvanitakis and N. J. Greenberger, Lancet, i (1976) 663. 

41. A. C. Bratton and E. K. Marshall, J. Biol. Chem., 128 (1939) 537. 

42. C. Yamato and K. Kinoshita, Anal. Biochem. , 98 (1979) 13. 

43. H. T. Karnes, R. D. Bateh, J. D. Winefordner and S. G. Schulman, 
submitted to Clin. Chem., May (1984). 

44. R. T. Parker, R. S. Friedlander, R. B. Dunlap, Anal. Chim. Acta., 
119 (1980) 189. 

45. T. Vo-Dinh, "Room Temperature Phosphorescence," Wiley 
Interscience, New York, 1984. 

46. G. N. Lewis and M. Kasha, J. Am. Chem. Soc, 66 (1944) 2100. 

47. R. T. Parker, R. S. Friedlander, E. M. Schulman, and R. B. 
Dunlap, Anal. Chem., 51 (1979) 1921. 

48. D. G. Peters, J. M. Hayes and G. M. Hiettie, Chemical Separations 
and Measurements, W. B. Saunders, Philadelphia, 1974. 

49. S. G. Schulman, "Fluorescence and Phosphorescence Spectroscopy: 
Physiochemical Principles and Practice," Pergamon Press, 
Elmsford, NY, 1977. 

50. S. K. Lower and M. A. El-Sayed, Chem. Rev., 66 (1966) 199. 

51. C. A. Parker and C. G. Hatchard, Analyst, 87 (1963) 664. 

52. J. J. Aaron and J. D. Winefordner, Talanta, 22 (1975) 707. 

53. M. Skrilee and L. J. Clinelove, Anal. Chem., 52 (1980) 1952. 

54. C. A. Parker, "Photoluminescence of Solutions," Elsevier, New 
York, 1968. 

55. P. Seybold and M. Gouterman, Chem. Rev., 65 (1965) 413. 

56. M. Zander, "Phosphorimetry , " Academic Press, New York, 1968, p. 
117. 

57. G. N. Lewis and M. Kasha, J. Am. Chem. Soc, 67 (1945) 994. 

58. N. A. Borisevich and A. A. Kotov, Spectrosc. Lett., 11 (1978) 
465. 



183 



59. P. Pringsheim, "Fluorescence and Phosphorescence," J. Wiley, New 
York, 1965. 

60. G. G. Guilbault, "Practical Fluorescence," M. Dekker, New York, 
1973. 

61. M. Roth, J. Chromatrogr. , 30 (1967) 276. 

62. E. M. Schulman and C. Walling, Science, 178 (1972) 53. 

63. S. L. Wellons, R. A. Paynther and J. D. Winefordner, Spectrochim. 
Acta., Part A, 30 (1974) 2133. 

64. E. H. Schulman and C. Walling, J. Phys. Chem., 77 (1973) 902. 

65. E. M. Schulman and R. T. Parker, J. Phys. Chem., 81 (1977) 1932. 

66. R. M. A. von Wandruszka and R. J. Hurtubise, Anal. Chem., 49 

(1977) 2164. 

67. R. M. A. von Wandruszka and R. J. Hurtubise, Anal. Chem., 48 
(1976) 1784. 

68. E. L. Y. Bower and J. D. Winefordner, Anal. Chim. Acta., 102 

(1978) 1. 

69. M. Kahsa, J. Chem. Phys., 20 (1952) 71. 

70. D. S. HcClure, J. Chem. Phys., 17 (1949) 905. 

71. E. H. Gilmore, G. E. Gibson and D. S. KcClure, J. Chem. Phys., 20 
(1952) 829. 

72. S. P. McGlynn, R. Sunseri and II. J. Christodouleas , J. Chem. 
Phys., 67 (1962) 1818. 

73. M. A. El-Sayed, Acc. Chem. Res., 1 (1968) 8. 

74. S. P. McGlynn, T. Asumi and M. Kinoshita, "Molecular Spectroscopy 
of the Triplet State," Prentice-Hall, Englewood Cliff, NJ, 1969. 

75. W. White and P. G. Seybold, J. Phys. Chem., 81 (1977) 2035. 

76. R. A. Paynther, S. L. Wellons and J. D. Winefordner, Anal. Chem.. 
46 (1974) 736. 

77. R. T. Parker, R. S. Freedlander and R. B. Dunlap, 1978, 
unpublished results. 

78. E. M. Schulman, J. Chem. Educ, 53 (1976) 522. 

79. T. Vo-Dinh and J. D. Winefordner, Appl. Spectrosc. Rev., 13 
(1977) 261. 



184 



80. G. J. Niday and P. G. Seybold, Anal. Chem. , 50 (1978) 1577. 

81. N. J. Turro, K. C. Lui, M. F. Chow and P. Lee, Photochem. 
Photobiol., 23 (1977) 523. 

82. L. J. Clinelove, M. Skrilee and J. G. Habarta, Anal. Chem., 52 
(1980) 754. 

83. R. T. Parker, R. S. Freedlander and R. B. Dunlap, Anal. Chim. 
Acta., 120 (1980) 1. 

84. E. L. Y. Bower, J. L. Ward, G. Walden and J. D. Winefordner, 
Talanta, 27 (1979) 380. 

85. R. J. Hurtubise, "Solid Surface Luminescence Analysis," M. 
Dekker, New York, 1981. 

86. R. A. Paynther, S. L. Wellons and J. D. Winefordner, Anal. Chem.. 
46 (1974) 736. 

87. E. M. Schulman and R. T. Parker, J. Phys. Chem., 81 (1977) 1932. 

88. T. Vo-Dinh and J. R. Hooyman, Anal. Chem., 51 (1979) 1915. 

89. C. D. Ford and R. J. Hurtubise, Anal. Chem., 51 (1979) 659. 

90. E. L. Y. Bower and J. D. Winefordner, Appl. Spectrosc, 33 (1979) 
9 • 

91. G. L. Walden and J. D. Winefordner, Appl. Spectrosc, 33 (1979) 
1 66 . 

92. T. Vo-Dinh and R. B. Gammage, Anal. Chem., 50 (1978) 2054. 

93. T. Vo-Dinh and R. B. Gammage, Anal. Chim. Acta., 107 (1979) 261. 

94. J. N. Miller, Trends Anal. Chem., 1 (1981) 31 . 

95. A. J. P. Martin and R. L. M. Synge, Biochem. J., 35 (1941 ) 91. 

96. R. A. Keller and G. H. Stewart, J. Chromatography, 9 (1962) 1. 

97. R. J. Hamilton and P. A. Sewell, "Introduction to High 
Performance Liquid Chromatography," Chapman and Hall, New York, 
1982. 

98. D. G. Peters, J. M. Hayes, and G. M. Hieftje, "Chemical 
Separations and Measurements," W. B. Saunders, Philadelphia, 
1974. 

'i 1 ^?" 1 ' 11611 ' " GaS Chromat °e ra P h y." John Wiley and Sons, New York, 



185 



100. J. C. Giddings, "Dynamics of Chromatography, Part I (Principles 
and Theory)," Dekker, New York, 1965. 

101. W. Feller, "Probability Theory and Its Applications," Wiley, New 
York, 1950. 

102. W. L. Jones, Anal. Chem. , 33 (1961 ) 829. 

103. J. C. Giddings, in "Chromatography," E. Heftmann, ed., Reinhold, 
New York, 1961 . 

104. J. C. Giddings, J. Chromatography, 2 (1959) 44. 

105. J. C. Giddings, J. Chem. Phys., 31 (1959) 1462. 

106. J. C. Giddings, Anal. Chem., 35 (1963) 439. 

107. J. C. Giddings, J. Chromatography, 3 (i960) 443. 
103. J. C. Giddings, J. Chromatography, 5 (1961 ) 46. 

109. J. C. Giddings, J. Phys. Chem., 68 (1964) 184. 

110. L. R. Snyder and J. J. Kirkland, "Introduction to Modern Liquid 
Chromatography," 2nd Ed., John Wiley and Sons, New York, 1979. 

111. K. K. Unger, N. Becker and P. Roumeliotis, J. Chromatogr. , 125 
(1976) 115. 



J. F. K. Huber, F. F. H. Kolder and J. M. Miller, Anal. Chem., 44 
(1972) 105. 



113. D. C. Locke, J. T. Schmermund and B. Banner, Anal. Chem., 44 
(1972) 90. 

114. I. Sebestian, 0. E. Brust and I. Halasz, in "Gas Chromatography," 
I. Montreux, S. G. Perry, ed., Applied Science Publishers, 
Barking, U. K. , 1972, p. 281. 

115. L. Boksanyi, 0. Liardon and E. sz. Kovats, J. Adv. Colloid 
Interface Sci., 6 (1976), 95. 

116. K. Karch, I. Sebestian and I. Halasz, J. Chromatogr., 122 (1976) 



H. Hemetsberger, W. Maasfeld and H. Ricken, Chromatographia . 9 
(1976) 303. 

C. Horvath, W. Melander and I. Molnar, J. Chromatogr., 125 (1976) 



186 



119. F. Riedo, M. Czencz, 0. Liardon and E. sz. Kovats, Helv. Chim. 
Acta., 61 (1978) 1912. 

120. J. J. Kirkland, J. Chromatogr. Sci., 9 ( 1 97 1 ) 206. 

121. R. W. Yost, L. S. Ettre and R. D. Coulon, "Practical Liquid 
Chromatography, An Introduction," Perkin-Elmer, Norwalk, CT, 
1980. 

122. B. L. Karger, J. R. Gant, A. Hartkopf and P. H. Weiner, J. 
Chromatogr., 128 (1976) 65. 

123. S. R. Bakalyar, R. Mcllwrick and E. Roggendorf, J. Chromatogr., 
142 (1977) 353. 

124. S. R. Bakalyar and R. A. Henry, J. Chromatogr., '126 (1976) 327. 

125. R. E. Majors, in "Bonded Stationary Phases in Chromatography," E, 
Grushka, ed., Ann Arbor Science Publishers, Ann Arbor, MI, 1974, 
p. 146. 

126. S. Eksborg, P. -0. Lagerstrom, R. Modin and G. Schill, J. 
Chromatogr., 83 (1973) 99. 

127. D. P. Wittmer, N. 0. Noeisle and V/. G. Haney, Anal. Chem. , 47 

(1975) 1422. 

128. K. -G. V/ahlund and U. Lund, J. Chromatogr., 122 (1976) 269. 

129. J. H. Knox and J. J. Jurand, J. Chromatogr., 110 (1975) 103. 

130. B. Fransson, K. -G. Wahlund and G. Schill, J. Chromatogr., 125 

(1976) 327. 

131. J. H. Knox and G. R. Laird, J. Chromatogr., 122 (1976) 17. 

132. B. -A. Persson and P. -0. Lagerstrom, J. Chromatogr., 122 (1976) 
305 . 

133. J. C. Kraak and J. F. K. Huber, J. Chromatogr., 102 (1974) 333. 

134. S. C. Su, A. V. Hartkopf and B. L. Karger, J. Chromatogr., 199 
(1976) 523. 

135. W. R. Melander, K. Kalqhatgi and C. Horvath, J. Chromatogr., 201 
(1980) 201. 



136, 



J. H. Knox and J. J. Jurand, J. Chromatogr., 103 (1975) 311. 



137. J. Crommen, B. Fransson and G. Schill, J. Chromatogr., 142 (1977) 
283. 



187 



138. T. Vo-Dinh, E. Lue Yen and J. D. Winefordner, Anal. Chen., 48 
(1976) 1186. 

139. R. M. A. von Wandnuszka and R. J. Hurtubise, Anal. Chen., 48 
(1976) 1784. 

140. J. L. Ward, Thesis, University of Florida, 1980. 

141. J. L. Ward, R. P. Eatch and J. D. V/inefordner, Analyst, 107 
(1982) 335. 

142. R. P. Batch and J. D. Winef ordner, Talanta, 29 (1982) 713. 

143. A. I. Vogel, "Practical Organic Chemistry," Wiley and Sons, New 
York, 1956, p. 577. 

144. R. C. Weast, "Handbook of Chemistry and Physics," 54th Edition, 
CRC Press, Cleveland, 1974, p. 334. 

145. R. M. Silverstein, G. C. Eassler and T. C. Morrill, 
"Spectrometric Identification of Organic Compounds," Wiley and 
Sons, New York, 1981 . 

146. R. K. Case, T. Scratcherd and R. P. A. Wynne, J. Physiol 
(London), 2120 (1970) 1. 

147. N. W. Tietz, "Fundamentals of Clinical Chemistry," W. B. 
Saunders, Philadelphia, 1976, p. 1092. 

148. M. J. Williamson, Adria Laboratories Inc., personal 
communication. 

149. R. G. P. Steel and J. H. Torrie, "Principles and Practice of 
Statistics," McGraw-Hill, New York, 1960, p. 74. 



150. 



R. G. P. Steel and J. H. Torrie, "Principles and Practice oi 
Statistics," McGraw-Hill, New York, 1960, p. 78. 



151. S. Y. Su and J. D. Winefordner, Can. J. Spectrosc, 26 (1983) 21 

152. W. L. White, M. M. Erickson and S. C. Stevens, "Chemistry for 
Medical Technologists," C.V. Mosby, St. Louis, 1970, p. 369. 

153. S. Ito, K. Moruta and Y. Inai, Clin. Chem. , 28 (1982) 323. 

154. R. W. Taft, E. Gurka and L. Joris, J. Am. Chem. Soc. 91 (196°) 
4801. 

155. C. R. Synder, J. Chromatogr. , 92 (1974) 223. 

156. L. R. Snyder and J. J. Kirkland, "Introduction to Modern Liauid 
Chromatography," 2nd Ed., Wiley Interscier.ee, New Yor> 1970 t, 



188 



157. P. J. Twrtchett and A. C. Moffat, J. Chromatogr. , 111 (1975) 149. 

158. D. If. Newton and R. B. Kluza, Drug Intel. & Clin. Pharm. , 12 
(1978) 546. 

159. E. Tomlinson, J. Pharmaceutical and Biomedical Analysis, 1 (1983) 

1 • 

160. E. Tomlinson and S. S. Davis, J. Colloid. Interface Sci., 74 

(1980) 349. 

161. S. C. Su, A. V. Hartkopf and B. C. Karger, J. Chromatogr., 199 
(1976) 523. 

162. C. M. Riley, E. Tomlinson and T. M. Jefferies, J. Chromatogr., 
185 (1979) 197. 

163. W. J. Conover, "Practical Nonparametric Statistics," Wiley and 
Sons, New York, 1971, p. 280. 

164. J. L. Ward, E. Lue Yen-Bower and J. D. Winefordner, Talanta, 28 

(1981) 119. 

165. S. G. Schulman, "Fluorescence and Phosphorescence Spectroscopy: 
Physicochemical Principles and Practice," Pergamon Press, New 
York, 1977, p. 42. 

166. M. A. El-Sayed, J. Chem. Phys., 36 (1962) 573. 

167. G. G. Giachino and D. R. Kearns, J. Chem. Phys., 54 (1971 ) 3248. 



BIOGRAPHICAL SKETCH 
H. Thomas Karnes was born in Cincinnati, Ohio, on September 27, 
1952. He was graduated from Peoria High School in Peoria, Illinois, in 
1970. He attended Illinois Central College in East Peoria, where he 
received an Associate of Arts degree majoring in medical technology in 
May, 1973. In August, 1974, he began study at Illinois State University 
where he received a Bachelor of Science degree in chemistry in May, 
1977. He entered the Graduate School at the University of Florida in 
September, 1977, where he served as a graduate assistant in the 
Department of Chemistry and subsequently in the Department of Pathology 
where he received a Master of Science degree majoring in clinical 
chemistry in December, 1980. From January, 1981, to August, 1982, he 
was employed by the Departments of Clinical Chemistry and Pharmacy 
Practice at the J. Hillis Miller Health Center with the responsibilities 
of developing an analytical toxicology laboratory and analytical support 
for pediatric pharmacokinetics, respectively. While pursuing his 
graduate education, he has published nine articles and one book chapter 
in the area of drug analysis. 

H. Thomas Karnes is married to Susan D. Hansen and has one child. 
He is a licensed medical technologist and a member of the American 
Pharmaceutical Association, the American Association of Pharmaceutical 
Sciences, the American Association for Clinical Chemistry and the 
American Chemical Society. Upon receiving his doctorate, he will join 



189 



190 



the faculty at the Medical College of Virginia in Richmond as an 
assistant professor. 



I certify that I have read this study and that in my opinion it 
conforms to acceptable standards of scholarly presentation and is fully 
adequate, in scope and quality, as a dissertation for the degree of 



Doctor of Philosophy. 



S. G. Schulman, Chairman 
Professor of Pharmacy 



I certify that I have read this study and that in my opinion it 
conforms to acceptable standards of scholarly presentation and is fully 
adequate, in scope and quality, as a dissertation for the degree of 
Doctor of Philosophy. 





Stephen H. Curry- 
Professor of Pharmacy 



I certify that I have read this study and that in my opinion it 
conforms to acceptable standards of scholarly presentation and is fully 
adequate, in scope and quality, as a dissertation for the degree of 
Doctor of Philosophy. 




Assistant Professor of Medicinal 
Chemistry 



I certify that I have read this study ana that in my opinion it 
conforms to acceptable standards of scholarly presentation and is fully 
adequate, in scope and quality, as a dissertation for the degree of 
Doctor of Philosophy. 



J. D. Winefordner 
Graduate Research Professor of 
Chemistry 



This dissertation was submitted to the Graduate Faculty of the College 
of Pharmacy and to the Graduate School, and was accepted as partial 
fulfillment of the requirements for the degree of Doctor of Philosophy. 

f t ■ — 

December 1984 ; '- '• L( - ' 



Dean, College of Pharmacy 



Dean for Graduate Studies 
Research 



and